From 8a50f79686839de58cf1ccde7c29da76b6c02224 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Thu, 17 Nov 2016 15:28:32 +0800 Subject: [PATCH 01/60] Convert markdown to rst for h-rnn --- doc_cn/algorithm/rnn/hierarchical-rnn.md | 403 -------------------- doc_cn/algorithm/rnn/hierarchical-rnn.rst | 440 ++++++++++++++++++++++ 2 files changed, 440 insertions(+), 403 deletions(-) delete mode 100644 doc_cn/algorithm/rnn/hierarchical-rnn.md create mode 100644 doc_cn/algorithm/rnn/hierarchical-rnn.rst diff --git a/doc_cn/algorithm/rnn/hierarchical-rnn.md b/doc_cn/algorithm/rnn/hierarchical-rnn.md deleted file mode 100644 index c184a34e85..0000000000 --- a/doc_cn/algorithm/rnn/hierarchical-rnn.md +++ /dev/null @@ -1,403 +0,0 @@ -# 双层RNN配置与示例 - -我们在`paddle/gserver/tests/test_RecurrentGradientMachine`单测中，通过多组语义相同的单双层RNN配置，讲解如何使用双层RNN。 - -## 示例1：双进双出，subseq间无memory - -配置：单层RNN（`sequence_layer_group`）和双层RNN（`sequence_nest_layer_group`），语义完全相同。 - -### 读取双层序列的方法 - -首先，我们看一下单双层序列的不同数据组织形式（您也可以采用别的组织形式）： - -- 单层序列的数据（`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 - -```text -2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 -2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * -2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 -2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 -2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . -2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 -2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ -2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 -2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * -2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 -``` - -- 双层序列的数据（`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 - -```text -2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 -2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * - -2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 -2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 -2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . - -2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 -2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ - -2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 -2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * -2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 -``` - -其次，我们看一下单双层序列的不同dataprovider（见`sequenceGen.py`）： - -- 单层序列的dataprovider如下： - - word_slot是integer_value_sequence类型，代表单层序列。 - - label是integer_value类型，代表一个向量。 - -```python -def hook(settings, dict_file, **kwargs): - settings.word_dict = dict_file - settings.input_types = [integer_value_sequence(len(settings.word_dict)), - integer_value(3)] - -@provider(init_hook=hook) -def process(settings, file_name): - with open(file_name, 'r') as fdata: - for line in fdata: - label, comment = line.strip().split('\t') - label = int(''.join(label.split())) - words = comment.split() - word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] - yield word_slot, label -``` - -- 双层序列的dataprovider如下： - - word_slot是integer_value_sub_sequence类型，代表双层序列。 - - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 - - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 - -```python -def hook2(settings, dict_file, **kwargs): - settings.word_dict = dict_file - settings.input_types = [integer_value_sub_sequence(len(settings.word_dict)), - integer_value_sequence(3)] - -@provider(init_hook=hook2) -def process2(settings, file_name): - with open(file_name) as fdata: - label_list = [] - word_slot_list = [] - for line in fdata: - if (len(line)) > 1: - label,comment = line.strip().split('\t') - label = int(''.join(label.split())) - words = comment.split() - word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] - label_list.append(label) - word_slot_list.append(word_slot) - else: - yield word_slot_list, label_list - label_list = [] - word_slot_list = [] -``` - -### 模型中的配置 - -首先，我们看一下单层序列的配置（见`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 - -```python -settings(batch_size=5) - -data = data_layer(name="word", size=dict_dim) - -emb = embedding_layer(input=data, size=word_dim) - -# (lstm_input + lstm) is equal to lstmemory -with mixed_layer(size=hidden_dim*4) as lstm_input: - lstm_input += full_matrix_projection(input=emb) - -lstm = lstmemory_group(input=lstm_input, - size=hidden_dim, - act=TanhActivation(), - gate_act=SigmoidActivation(), - state_act=TanhActivation(), - lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) - -lstm_last = last_seq(input=lstm) - -with mixed_layer(size=label_dim, - act=SoftmaxActivation(), - bias_attr=True) as output: - output += full_matrix_projection(input=lstm_last) - -outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) - -``` -其次，我们看一下语义相同的双层序列配置（见`sequence_nest_layer_group.conf`），并对其详细分析： - -- batchsize=2表示一次过2句双层序列。但从上面的数据格式可知，2句双层序列和5句单层序列的数据完全一样。 -- data_layer和embedding_layer不关心数据是否是序列格式，因此两个配置在这两层上的输出是一样的。 -- lstmemory: - - 单层序列过了一个mixed_layer和lstmemory_group。 - - 双层序列在同样的mixed_layer和lstmemory_group外，直接加了一层group。由于这个外层group里面没有memory，表示subseq间不存在联系，即起到的作用仅仅是把双层seq拆成单层，因此双层序列过完lstmemory的输出和单层的一样。 -- last_seq： - - 单层序列直接取了最后一个元素 - - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 - - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 - -```python -settings(batch_size=2) - -data = data_layer(name="word", size=dict_dim) - -emb_group = embedding_layer(input=data, size=word_dim) - -# (lstm_input + lstm) is equal to lstmemory -def lstm_group(lstm_group_input): - with mixed_layer(size=hidden_dim*4) as group_input: - group_input += full_matrix_projection(input=lstm_group_input) - - lstm_output = lstmemory_group(input=group_input, - name="lstm_group", - size=hidden_dim, - act=TanhActivation(), - gate_act=SigmoidActivation(), - state_act=TanhActivation(), - lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) - return lstm_output - -lstm_nest_group = recurrent_group(input=SubsequenceInput(emb_group), - step=lstm_group, - name="lstm_nest_group") -# hasSubseq ->(seqlastins) seq -lstm_last = last_seq(input=lstm_nest_group, agg_level=AggregateLevel.EACH_SEQUENCE) - -# seq ->(expand) hasSubseq -lstm_expand = expand_layer(input=lstm_last, expand_as=emb_group, expand_level=ExpandLevel.FROM_SEQUENCE) - -# hasSubseq ->(average) seq -lstm_average = pooling_layer(input=lstm_expand, - pooling_type=AvgPooling(), - agg_level=AggregateLevel.EACH_SEQUENCE) - -with mixed_layer(size=label_dim, - act=SoftmaxActivation(), - bias_attr=True) as output: - output += full_matrix_projection(input=lstm_average) - -outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) -``` -## 示例2：双进双出，subseq间有memory - -配置：单层RNN（`sequence_rnn.conf`），双层RNN（`sequence_nest_rnn.conf`和`sequence_nest_rnn_readonly_memory.conf`），语义完全相同。 - -### 读取双层序列的方法 - -我们看一下单双层序列的不同数据组织形式和dataprovider（见`rnn_data_provider.py`） -```python -data = [ - [[[1, 3, 2], [4, 5, 2]], 0], - [[[0, 2], [2, 5], [0, 1, 2]], 1], -] - -@provider(input_types=[integer_value_sub_sequence(10), - integer_value(3)]) -def process_subseq(settings, file_name): - for d in data: - yield d - -@provider(input_types=[integer_value_sequence(10), - integer_value(3)]) -def process_seq(settings, file_name): - for d in data: - seq = [] -``` -- 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 -- 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 -- 单双层序列的label都分别是0和1 - -### 模型中的配置 - -我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 - -- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 - -```python -def step(y): - mem = memory(name="rnn_state", size=hidden_dim) - return fc_layer(input=[y, mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name="rnn_state") - -out = recurrent_group(step=step, input=emb) -``` -- 双层序列，外层memory是一个元素： - - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 - -```python -def outer_step(x): - outer_mem = memory(name="outer_rnn_state", size=hidden_dim) - def inner_step(y): - inner_mem = memory(name="inner_rnn_state", - size=hidden_dim, - boot_layer=outer_mem) - return fc_layer(input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name="inner_rnn_state") - - inner_rnn_output = recurrent_group( - step=inner_step, - input=x) - last = last_seq(input=inner_rnn_output, name="outer_rnn_state") - - return inner_rnn_output - -out = recurrent_group(step=outer_step, input=SubsequenceInput(emb)) -``` -- 双层序列，外层memory是单层序列： - - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 - - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 - -## 示例3：双进双出，输入不等长 - -**输入不等长**是指recurrent_group的多个输入在各时刻的长度可以不相等, 但需要指定一个和输出长度一致的input，用targetInlink表示。参考配置：单层RNN（`sequence_rnn_multi_unequalength_inputs.conf`），双层RNN（`sequence_nest_rnn_multi_unequalength_inputs.conf`） - -### 读取双层序列的方法 - -我们看一下单双层序列的数据组织形式和dataprovider（见`rnn_data_provider.py`） -```python -data2 = [ - [[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]] ,0], - [[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]], 1], -] - -@provider(input_types=[integer_value_sub_sequence(10), - integer_value_sub_sequence(10), - integer_value(2)], - should_shuffle=False) -def process_unequalength_subseq(settings, file_name): #双层RNN的dataprovider - for d in data2: - yield d - - -@provider(input_types=[integer_value_sequence(10), - integer_value_sequence(10), - integer_value(2)], - should_shuffle=False) -def process_unequalength_seq(settings, file_name): #单层RNN的dataprovider - for d in data2: - words1=reduce(lambda x,y: x+y, d[0]) - words2=reduce(lambda x,y: x+y, d[1]) - yield words1, words2, d[2] -``` - -data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 - -- 单层序列：两个样本分别为[[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]] 和 [[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]] -- 双层序列：两个样本分别为 - - **样本1**：[[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]]]。fea1和fea2都分别有2个子句，fea1=[[1, 2], [4, 5, 2]], fea2=[[5, 4, 1], [3, 1]] - - **样本2**：[[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]]。fea1和fea2都分别有3个子句， fea1=[[0, 2], [2, 5], [0, 1, 2]], fea2=[[1, 5], [4], [2, 3, 6, 1]]。
- - **注意**：每个样本中，各特征的子句数目需要相等。这里说的“双进双出，输入不等长”是指fea1在i时刻的输入的长度可以不等于fea2在i时刻的输入的长度。如对于第1个样本，时刻i=2, fea1[2]=[4, 5, 2]，fea2[2]=[3, 1]，3≠2。 -- 单双层序列中，两个样本的label都分别是0和1 - -### 模型中的配置 - -单层RNN（`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（`sequence_nest_rnn_multi_unequalength_inputs.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 - -- 单层序列： - - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 - -```python -def step(x1, x2): - def calrnn(y): - mem = memory(name = 'rnn_state_' + y.name, size = hidden_dim) - out = fc_layer(input = [y, mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'rnn_state_' + y.name) - return out - - encoder1 = calrnn(x1) - encoder2 = calrnn(x2) - return [encoder1, encoder2] - -encoder1_rep, encoder2_rep = recurrent_group( - name="stepout", - step=step, - input=[emb1, emb2]) - -encoder1_last = last_seq(input = encoder1_rep) -encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) -context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) -``` -- 双层序列： - - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 - - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - -```python -def outer_step(x1, x2): - outer_mem1 = memory(name = "outer_rnn_state1", size = hidden_dim) - outer_mem2 = memory(name = "outer_rnn_state2", size = hidden_dim) - def inner_step1(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem1) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - def inner_step2(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem2) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - encoder1 = recurrent_group( - step = inner_step1, - name = 'inner1', - input = x1) - - encoder2 = recurrent_group( - step = inner_step2, - name = 'inner2', - input = x2) - - sentence_last_state1 = last_seq(input = encoder1, name = 'outer_rnn_state1') - sentence_last_state2_ = last_seq(input = encoder2, name = 'outer_rnn_state2') - - encoder1_expand = expand_layer(input = sentence_last_state1, - expand_as = encoder2) - - return [encoder1_expand, encoder2] - -encoder1_rep, encoder2_rep = recurrent_group( - name="outer", - step=outer_step, - input=[SubsequenceInput(emb1), SubsequenceInput(emb2)], - targetInlink=emb2) - -encoder1_last = last_seq(input = encoder1_rep) -encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) -context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) -``` - -## 示例4：beam_search的生成 - -TBD diff --git a/doc_cn/algorithm/rnn/hierarchical-rnn.rst b/doc_cn/algorithm/rnn/hierarchical-rnn.rst new file mode 100644 index 0000000000..7de54cc0b1 --- /dev/null +++ b/doc_cn/algorithm/rnn/hierarchical-rnn.rst @@ -0,0 +1,440 @@ +################# +双层RNN配置与示例 +################# + +我们在 :code:`paddle/gserver/tests/test_RecurrentGradientMachine` 单测中，通过多组语义相同的单双层RNN配置，讲解如何使用双层RNN。 + +示例1：双进双出，subseq间无memory +================================= + +配置：单层RNN（:code:`sequence_layer_group`）和双层RNN（:code:`sequence_nest_layer_group`），语义完全相同。 + +读取双层序列的方法 +------------------ + +首先，我们看一下单双层序列的不同数据组织形式（您也可以采用别的组织形式）\: + +- 单层序列的数据（ :code:`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 + +.. code-block:: text + + 2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 + 2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * + 2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 + 2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 + 2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . + 2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 + 2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ + 2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 + 2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * + 2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 + + +- 双层序列的数据（ :code:`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 + +.. code-block:: text + + 2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 + 2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * + + 2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 + 2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 + 2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . + + 2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 + 2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ + + 2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 + 2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * + 2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 + +其次，我们看一下单双层序列的不同dataprovider（见 :code:`sequenceGen.py` ）： + +- 单层序列的dataprovider如下： + + - word_slot是integer_value_sequence类型，代表单层序列。 + - label是integer_value类型，代表一个向量。 + +.. code-block:: python + + def hook(settings, dict_file, **kwargs): + settings.word_dict = dict_file + settings.input_types = [integer_value_sequence(len(settings.word_dict)), + integer_value(3)] + + @provider(init_hook=hook) + def process(settings, file_name): + with open(file_name, 'r') as fdata: + for line in fdata: + label, comment = line.strip().split('\t') + label = int(''.join(label.split())) + words = comment.split() + word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] + yield word_slot, label + +- 双层序列的dataprovider如下： + + - word_slot是integer_value_sub_sequence类型，代表双层序列。 + - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 + - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 + +.. code-block:: python + + def hook2(settings, dict_file, **kwargs): + settings.word_dict = dict_file + settings.input_types = [integer_value_sub_sequence(len(settings.word_dict)), + integer_value_sequence(3)] + + @provider(init_hook=hook2) + def process2(settings, file_name): + with open(file_name) as fdata: + label_list = [] + word_slot_list = [] + for line in fdata: + if (len(line)) > 1: + label,comment = line.strip().split('\t') + label = int(''.join(label.split())) + words = comment.split() + word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] + label_list.append(label) + word_slot_list.append(word_slot) + else: + yield word_slot_list, label_list + label_list = [] + word_slot_list = [] + + +模型中的配置 +------------ + +首先，我们看一下单层序列的配置（见 :code:`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 + +.. code-block:: python + + settings(batch_size=5) + + data = data_layer(name="word", size=dict_dim) + + emb = embedding_layer(input=data, size=word_dim) + + # (lstm_input + lstm) is equal to lstmemory + with mixed_layer(size=hidden_dim*4) as lstm_input: + lstm_input += full_matrix_projection(input=emb) + + lstm = lstmemory_group(input=lstm_input, + size=hidden_dim, + act=TanhActivation(), + gate_act=SigmoidActivation(), + state_act=TanhActivation(), + lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) + + lstm_last = last_seq(input=lstm) + + with mixed_layer(size=label_dim, + act=SoftmaxActivation(), + bias_attr=True) as output: + output += full_matrix_projection(input=lstm_last) + + outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) + + +其次，我们看一下语义相同的双层序列配置（见 :code:`sequence_nest_layer_group.conf` ），并对其详细分析： + +- batchsize=2表示一次过2句双层序列。但从上面的数据格式可知，2句双层序列和5句单层序列的数据完全一样。 +- data_layer和embedding_layer不关心数据是否是序列格式，因此两个配置在这两层上的输出是一样的。 +- lstmemory\: + + - 单层序列过了一个mixed_layer和lstmemory_group。 + - 双层序列在同样的mixed_layer和lstmemory_group外，直接加了一层group。由于这个外层group里面没有memory，表示subseq间不存在联系，即起到的作用仅仅是把双层seq拆成单层，因此双层序列过完lstmemory的输出和单层的一样。 + +- last_seq\: + + - 单层序列直接取了最后一个元素 + - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 + - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 + +.. code-block:: python + + settings(batch_size=2) + + data = data_layer(name="word", size=dict_dim) + + emb_group = embedding_layer(input=data, size=word_dim) + + # (lstm_input + lstm) is equal to lstmemory + def lstm_group(lstm_group_input): + with mixed_layer(size=hidden_dim*4) as group_input: + group_input += full_matrix_projection(input=lstm_group_input) + + lstm_output = lstmemory_group(input=group_input, + name="lstm_group", + size=hidden_dim, + act=TanhActivation(), + gate_act=SigmoidActivation(), + state_act=TanhActivation(), + lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) + return lstm_output + + lstm_nest_group = recurrent_group(input=SubsequenceInput(emb_group), + step=lstm_group, + name="lstm_nest_group") + # hasSubseq ->(seqlastins) seq + lstm_last = last_seq(input=lstm_nest_group, agg_level=AggregateLevel.EACH_SEQUENCE) + + # seq ->(expand) hasSubseq + lstm_expand = expand_layer(input=lstm_last, expand_as=emb_group, expand_level=ExpandLevel.FROM_SEQUENCE) + + # hasSubseq ->(average) seq + lstm_average = pooling_layer(input=lstm_expand, + pooling_type=AvgPooling(), + agg_level=AggregateLevel.EACH_SEQUENCE) + + with mixed_layer(size=label_dim, + act=SoftmaxActivation(), + bias_attr=True) as output: + output += full_matrix_projection(input=lstm_average) + + outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) + +示例2：双进双出，subseq间有memory +================================= + +配置：单层RNN（ :code:`sequence_rnn.conf` ），双层RNN（ :code:`sequence_nest_rnn.conf` 和 :code:`sequence_nest_rnn_readonly_memory.conf` ），语义完全相同。 + +读取双层序列的方法 +------------------ + +我们看一下单双层序列的不同数据组织形式和dataprovider（见 :code:`rnn_data_provider.py`） + +.. code-block:: python + + data = [ + [[[1, 3, 2], [4, 5, 2]], 0], + [[[0, 2], [2, 5], [0, 1, 2]], 1], + ] + + @provider(input_types=[integer_value_sub_sequence(10), + integer_value(3)]) + def process_subseq(settings, file_name): + for d in data: + yield d + + @provider(input_types=[integer_value_sequence(10), + integer_value(3)]) + def process_seq(settings, file_name): + for d in data: + seq = [] + +- 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 +- 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 +- 单双层序列的label都分别是0和1 + +模型中的配置 +------------ + +我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 + +- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 + +.. code-block:: python + + def step(y): + mem = memory(name="rnn_state", size=hidden_dim) + return fc_layer(input=[y, mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name="rnn_state") + + out = recurrent_group(step=step, input=emb) + +- 双层序列，外层memory是一个元素： + - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 + - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 + +.. code-block:: + + def outer_step(x): + outer_mem = memory(name="outer_rnn_state", size=hidden_dim) + def inner_step(y): + inner_mem = memory(name="inner_rnn_state", + size=hidden_dim, + boot_layer=outer_mem) + return fc_layer(input=[y, inner_mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name="inner_rnn_state") + + inner_rnn_output = recurrent_group( + step=inner_step, + input=x) + last = last_seq(input=inner_rnn_output, name="outer_rnn_state") + + return inner_rnn_output + + out = recurrent_group(step=outer_step, input=SubsequenceInput(emb)) + +- 双层序列，外层memory是单层序列： + - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 + - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 + +示例3：双进双出，输入不等长 +=========================== + +.. role:: red + +.. raw:: html + + + +**输入不等长** 是指recurrent_group的多个输入在各时刻的长度可以不相等, 但需要指定一个和输出长度一致的input，用 :red:`targetInlink` 表示。参考配置：单层RNN（:code:`sequence_rnn_multi_unequalength_inputs.conf`），双层RNN（:code:`sequence_nest_rnn_multi_unequalength_inputs.conf`） + +读取双层序列的方法 +------------------ + +我们看一下单双层序列的数据组织形式和dataprovider（见`rnn_data_provider.py`） + +.. code-block:: python + + data2 = [ + [[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]] ,0], + [[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]], 1], + ] + + @provider(input_types=[integer_value_sub_sequence(10), + integer_value_sub_sequence(10), + integer_value(2)], + should_shuffle=False) + def process_unequalength_subseq(settings, file_name): #双层RNN的dataprovider + for d in data2: + yield d + + + @provider(input_types=[integer_value_sequence(10), + integer_value_sequence(10), + integer_value(2)], + should_shuffle=False) + def process_unequalength_seq(settings, file_name): #单层RNN的dataprovider + for d in data2: + words1=reduce(lambda x,y: x+y, d[0]) + words2=reduce(lambda x,y: x+y, d[1]) + yield words1, words2, d[2] + +data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 + +- 单层序列：两个样本分别为[[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]] 和 [[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]] +- 双层序列：两个样本分别为 + + - **样本1**\：[[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]]]。fea1和fea2都分别有2个子句，fea1=[[1, 2], [4, 5, 2]], fea2=[[5, 4, 1], [3, 1]] + - **样本2**\：[[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]]。fea1和fea2都分别有3个子句， fea1=[[0, 2], [2, 5], [0, 1, 2]], fea2=[[1, 5], [4], [2, 3, 6, 1]]。
+ - **注意**\：每个样本中，各特征的子句数目需要相等。这里说的“双进双出，输入不等长”是指fea1在i时刻的输入的长度可以不等于fea2在i时刻的输入的长度。如对于第1个样本，时刻i=2, fea1[2]=[4, 5, 2]，fea2[2]=[3, 1]，3≠2。 + +- 单双层序列中，两个样本的label都分别是0和1 + +模型中的配置 +------------ + +单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`sequence_nest_rnn_multi_unequalength_inputs.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 + +- 单层序列\： + + - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 + - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 + +.. code-block:: python + + def step(x1, x2): + def calrnn(y): + mem = memory(name = 'rnn_state_' + y.name, size = hidden_dim) + out = fc_layer(input = [y, mem], + size = hidden_dim, + act = TanhActivation(), + bias_attr = True, + name = 'rnn_state_' + y.name) + return out + + encoder1 = calrnn(x1) + encoder2 = calrnn(x2) + return [encoder1, encoder2] + + encoder1_rep, encoder2_rep = recurrent_group( + name="stepout", + step=step, + input=[emb1, emb2]) + + encoder1_last = last_seq(input = encoder1_rep) + encoder1_expandlast = expand_layer(input = encoder1_last, + expand_as = encoder2_rep) + context = mixed_layer(input = [identity_projection(encoder1_expandlast), + identity_projection(encoder2_rep)], + size = hidden_dim) + +- 双层序列\： + + - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 + - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 + - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 + +.. code-block:: python + + def outer_step(x1, x2): + outer_mem1 = memory(name = "outer_rnn_state1", size = hidden_dim) + outer_mem2 = memory(name = "outer_rnn_state2", size = hidden_dim) + def inner_step1(y): + inner_mem = memory(name = 'inner_rnn_state_' + y.name, + size = hidden_dim, + boot_layer = outer_mem1) + out = fc_layer(input = [y, inner_mem], + size = hidden_dim, + act = TanhActivation(), + bias_attr = True, + name = 'inner_rnn_state_' + y.name) + return out + + def inner_step2(y): + inner_mem = memory(name = 'inner_rnn_state_' + y.name, + size = hidden_dim, + boot_layer = outer_mem2) + out = fc_layer(input = [y, inner_mem], + size = hidden_dim, + act = TanhActivation(), + bias_attr = True, + name = 'inner_rnn_state_' + y.name) + return out + + encoder1 = recurrent_group( + step = inner_step1, + name = 'inner1', + input = x1) + + encoder2 = recurrent_group( + step = inner_step2, + name = 'inner2', + input = x2) + + sentence_last_state1 = last_seq(input = encoder1, name = 'outer_rnn_state1') + sentence_last_state2_ = last_seq(input = encoder2, name = 'outer_rnn_state2') + + encoder1_expand = expand_layer(input = sentence_last_state1, + expand_as = encoder2) + + return [encoder1_expand, encoder2] + + encoder1_rep, encoder2_rep = recurrent_group( + name="outer", + step=outer_step, + input=[SubsequenceInput(emb1), SubsequenceInput(emb2)], + targetInlink=emb2) + + encoder1_last = last_seq(input = encoder1_rep) + encoder1_expandlast = expand_layer(input = encoder1_last, + expand_as = encoder2_rep) + context = mixed_layer(input = [identity_projection(encoder1_expandlast), + identity_projection(encoder2_rep)], + size = hidden_dim) + +示例4：beam_search的生成 +======================== + +TBD From f1955e2b2092d63f9acab07d8971ef2c5572170c Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Thu, 17 Nov 2016 16:36:26 +0800 Subject: [PATCH 02/60] Remove copy & paste code. --- doc_cn/algorithm/rnn/hierarchical-rnn.rst | 337 +++------------------- 1 file changed, 38 insertions(+), 299 deletions(-) diff --git a/doc_cn/algorithm/rnn/hierarchical-rnn.rst b/doc_cn/algorithm/rnn/hierarchical-rnn.rst index 7de54cc0b1..7c81ce8c67 100644 --- a/doc_cn/algorithm/rnn/hierarchical-rnn.rst +++ b/doc_cn/algorithm/rnn/hierarchical-rnn.rst @@ -16,37 +16,14 @@ - 单层序列的数据（ :code:`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 -.. code-block:: text - - 2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 - 2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * - 2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 - 2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 - 2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . - 2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 - 2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ - 2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 - 2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * - 2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 +.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg + :language: text - 双层序列的数据（ :code:`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 -.. code-block:: text - - 2 酒店 有 很 舒适 的 床垫 子 ， 床上用品 也 应该 是 一人 一 换 ， 感觉 很 利落 对 卫生 很 放心 呀 。 - 2 很 温馨 ， 也 挺 干净 的 * 地段 不错 ， 出来 就 有 全家 ， 离 地铁站 也 近 ， 交通 很方便 * 就是 都 不 给 刷牙 的 杯子 啊 ， 就 第一天 给 了 一次性杯子 * - - 2 位置 方便 ， 强烈推荐 ， 十一 出去玩 的 时候 选 的 ， 对面 就是 华润万家 ， 周围 吃饭 的 也 不少 。 - 2 交通便利 ， 吃 很 便利 ， 乾 浄 、 安静 ， 商务 房 有 电脑 、 上网 快 ， 价格 可以 ， 就 早餐 不 好吃 。 整体 是 不错 的 。 適 合 出差 來 住 。 - 2 本来 准备 住 两 晚 ， 第 2 天 一早 居然 停电 ， 且 无 通知 ， 只有 口头 道歉 。 总体来说 性价比 尚可 ， 房间 较 新 ， 还是 推荐 . - - 2 这个 酒店 去过 很多 次 了 ， 选择 的 主要原因 是 离 客户 最 便宜 相对 又 近 的 酒店 - 2 挺好 的 汉庭 ， 前台 服务 很 热情 ， 卫生 很 整洁 ， 房间 安静 ， 水温 适中 ， 挺好 ！ - - 2 HowardJohnson 的 品质 ， 服务 相当 好 的 一 家 五星级 。 房间 不错 、 泳池 不错 、 楼层 安排 很 合理 。 还有 就是 地理位置 ， 简直 一 流 。 就 在 天一阁 、 月湖 旁边 ， 离 天一广场 也 不远 。 下次 来 宁波 还会 住 。 - 2 酒店 很干净 ， 很安静 ， 很 温馨 ， 服务员 服务 好 ， 各方面 都 不错 * - 2 挺好 的 ， 就是 没 窗户 ， 不过 对 得 起 这 价格 +.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest + :language: text 其次，我们看一下单双层序列的不同dataprovider（见 :code:`sequenceGen.py` ）： @@ -55,22 +32,9 @@ - word_slot是integer_value_sequence类型，代表单层序列。 - label是integer_value类型，代表一个向量。 -.. code-block:: python - - def hook(settings, dict_file, **kwargs): - settings.word_dict = dict_file - settings.input_types = [integer_value_sequence(len(settings.word_dict)), - integer_value(3)] - - @provider(init_hook=hook) - def process(settings, file_name): - with open(file_name, 'r') as fdata: - for line in fdata: - label, comment = line.strip().split('\t') - label = int(''.join(label.split())) - words = comment.split() - word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] - yield word_slot, label +.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py + :language: python + :lines: 21-39 - 双层序列的dataprovider如下： @@ -78,64 +42,18 @@ - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 -.. code-block:: python - - def hook2(settings, dict_file, **kwargs): - settings.word_dict = dict_file - settings.input_types = [integer_value_sub_sequence(len(settings.word_dict)), - integer_value_sequence(3)] - - @provider(init_hook=hook2) - def process2(settings, file_name): - with open(file_name) as fdata: - label_list = [] - word_slot_list = [] - for line in fdata: - if (len(line)) > 1: - label,comment = line.strip().split('\t') - label = int(''.join(label.split())) - words = comment.split() - word_slot = [settings.word_dict[w] for w in words if w in settings.word_dict] - label_list.append(label) - word_slot_list.append(word_slot) - else: - yield word_slot_list, label_list - label_list = [] - word_slot_list = [] - +.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py + :language: python + :lines: 42-71 模型中的配置 ------------ 首先，我们看一下单层序列的配置（见 :code:`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 -.. code-block:: python - - settings(batch_size=5) - - data = data_layer(name="word", size=dict_dim) - - emb = embedding_layer(input=data, size=word_dim) - - # (lstm_input + lstm) is equal to lstmemory - with mixed_layer(size=hidden_dim*4) as lstm_input: - lstm_input += full_matrix_projection(input=emb) - - lstm = lstmemory_group(input=lstm_input, - size=hidden_dim, - act=TanhActivation(), - gate_act=SigmoidActivation(), - state_act=TanhActivation(), - lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) - - lstm_last = last_seq(input=lstm) - - with mixed_layer(size=label_dim, - act=SoftmaxActivation(), - bias_attr=True) as output: - output += full_matrix_projection(input=lstm_last) - - outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf + :language: python + :lines: 38-63 其次，我们看一下语义相同的双层序列配置（见 :code:`sequence_nest_layer_group.conf` ），并对其详细分析： @@ -153,48 +71,9 @@ - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 -.. code-block:: python - - settings(batch_size=2) - - data = data_layer(name="word", size=dict_dim) - - emb_group = embedding_layer(input=data, size=word_dim) - - # (lstm_input + lstm) is equal to lstmemory - def lstm_group(lstm_group_input): - with mixed_layer(size=hidden_dim*4) as group_input: - group_input += full_matrix_projection(input=lstm_group_input) - - lstm_output = lstmemory_group(input=group_input, - name="lstm_group", - size=hidden_dim, - act=TanhActivation(), - gate_act=SigmoidActivation(), - state_act=TanhActivation(), - lstm_layer_attr=ExtraLayerAttribute(error_clipping_threshold=50)) - return lstm_output - - lstm_nest_group = recurrent_group(input=SubsequenceInput(emb_group), - step=lstm_group, - name="lstm_nest_group") - # hasSubseq ->(seqlastins) seq - lstm_last = last_seq(input=lstm_nest_group, agg_level=AggregateLevel.EACH_SEQUENCE) - - # seq ->(expand) hasSubseq - lstm_expand = expand_layer(input=lstm_last, expand_as=emb_group, expand_level=ExpandLevel.FROM_SEQUENCE) - - # hasSubseq ->(average) seq - lstm_average = pooling_layer(input=lstm_expand, - pooling_type=AvgPooling(), - agg_level=AggregateLevel.EACH_SEQUENCE) - - with mixed_layer(size=label_dim, - act=SoftmaxActivation(), - bias_attr=True) as output: - output += full_matrix_projection(input=lstm_average) - - outputs(classification_cost(input=output, label=data_layer(name="label", size=1))) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf + :language: python + :lines: 38-84 示例2：双进双出，subseq间有memory ================================= @@ -206,24 +85,9 @@ 我们看一下单双层序列的不同数据组织形式和dataprovider（见 :code:`rnn_data_provider.py`） -.. code-block:: python - - data = [ - [[[1, 3, 2], [4, 5, 2]], 0], - [[[0, 2], [2, 5], [0, 1, 2]], 1], - ] - - @provider(input_types=[integer_value_sub_sequence(10), - integer_value(3)]) - def process_subseq(settings, file_name): - for d in data: - yield d - - @provider(input_types=[integer_value_sequence(10), - integer_value(3)]) - def process_seq(settings, file_name): - for d in data: - seq = [] +.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py + :language: python + :lines: 20-32 - 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 - 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 @@ -236,46 +100,21 @@ - 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 -.. code-block:: python - - def step(y): - mem = memory(name="rnn_state", size=hidden_dim) - return fc_layer(input=[y, mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name="rnn_state") - - out = recurrent_group(step=step, input=emb) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf + :language: python + :lines: 36-48 - 双层序列，外层memory是一个元素： + - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 -.. code-block:: - - def outer_step(x): - outer_mem = memory(name="outer_rnn_state", size=hidden_dim) - def inner_step(y): - inner_mem = memory(name="inner_rnn_state", - size=hidden_dim, - boot_layer=outer_mem) - return fc_layer(input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name="inner_rnn_state") - - inner_rnn_output = recurrent_group( - step=inner_step, - input=x) - last = last_seq(input=inner_rnn_output, name="outer_rnn_state") - - return inner_rnn_output - - out = recurrent_group(step=outer_step, input=SubsequenceInput(emb)) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf + :language: python + :lines: 39-66 - 双层序列，外层memory是单层序列： + - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 @@ -293,33 +132,11 @@ 读取双层序列的方法 ------------------ -我们看一下单双层序列的数据组织形式和dataprovider（见`rnn_data_provider.py`） - -.. code-block:: python - - data2 = [ - [[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]] ,0], - [[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]], 1], - ] - - @provider(input_types=[integer_value_sub_sequence(10), - integer_value_sub_sequence(10), - integer_value(2)], - should_shuffle=False) - def process_unequalength_subseq(settings, file_name): #双层RNN的dataprovider - for d in data2: - yield d - +我们看一下单双层序列的数据组织形式和dataprovider（见 :code:`rnn_data_provider.py` ） - @provider(input_types=[integer_value_sequence(10), - integer_value_sequence(10), - integer_value(2)], - should_shuffle=False) - def process_unequalength_seq(settings, file_name): #单层RNN的dataprovider - for d in data2: - words1=reduce(lambda x,y: x+y, d[0]) - words2=reduce(lambda x,y: x+y, d[1]) - yield words1, words2, d[2] +.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py + :language: python + :lines: 69-97 data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 @@ -335,40 +152,16 @@ data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 模型中的配置 ------------ -单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`sequence_nest_rnn_multi_unequalength_inputs.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 +单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`v.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 - 单层序列\： - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 -.. code-block:: python - - def step(x1, x2): - def calrnn(y): - mem = memory(name = 'rnn_state_' + y.name, size = hidden_dim) - out = fc_layer(input = [y, mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'rnn_state_' + y.name) - return out - - encoder1 = calrnn(x1) - encoder2 = calrnn(x2) - return [encoder1, encoder2] - - encoder1_rep, encoder2_rep = recurrent_group( - name="stepout", - step=step, - input=[emb1, emb2]) - - encoder1_last = last_seq(input = encoder1_rep) - encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) - context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf + :language: python + :lines: 41-58 - 双层序列\： @@ -376,63 +169,9 @@ data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 -.. code-block:: python - - def outer_step(x1, x2): - outer_mem1 = memory(name = "outer_rnn_state1", size = hidden_dim) - outer_mem2 = memory(name = "outer_rnn_state2", size = hidden_dim) - def inner_step1(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem1) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - def inner_step2(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem2) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - encoder1 = recurrent_group( - step = inner_step1, - name = 'inner1', - input = x1) - - encoder2 = recurrent_group( - step = inner_step2, - name = 'inner2', - input = x2) - - sentence_last_state1 = last_seq(input = encoder1, name = 'outer_rnn_state1') - sentence_last_state2_ = last_seq(input = encoder2, name = 'outer_rnn_state2') - - encoder1_expand = expand_layer(input = sentence_last_state1, - expand_as = encoder2) - - return [encoder1_expand, encoder2] - - encoder1_rep, encoder2_rep = recurrent_group( - name="outer", - step=outer_step, - input=[SubsequenceInput(emb1), SubsequenceInput(emb2)], - targetInlink=emb2) - - encoder1_last = last_seq(input = encoder1_rep) - encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) - context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf + :language: python + :lines: 41-89 示例4：beam_search的生成 ======================== From b3dd2d10b74b495aa525ec1052e5ed1b761aa935 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Fri, 18 Nov 2016 17:07:06 +0800 Subject: [PATCH 03/60] Add glossary for Paddle --- doc_cn/algorithm/rnn/hrnn_demo.rst | 7 + doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 183 ++++++++++++++++++ doc_cn/concepts/glossary.rst | 59 ++++++ doc_cn/concepts/glossary_rnn.dot | 42 ++++ doc_cn/concepts/glossary_rnn_with_memory.dot | 48 +++++ doc_cn/index.rst | 3 +- 6 files changed, 341 insertions(+), 1 deletion(-) create mode 100644 doc_cn/algorithm/rnn/hrnn_demo.rst create mode 100644 doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst create mode 100644 doc_cn/concepts/glossary.rst create mode 100644 doc_cn/concepts/glossary_rnn.dot create mode 100644 doc_cn/concepts/glossary_rnn_with_memory.dot diff --git a/doc_cn/algorithm/rnn/hrnn_demo.rst b/doc_cn/algorithm/rnn/hrnn_demo.rst new file mode 100644 index 0000000000..cf38e416c0 --- /dev/null +++ b/doc_cn/algorithm/rnn/hrnn_demo.rst @@ -0,0 +1,7 @@ +.. algo_hrnn_demo: + +################# +双层RNN的使用示例 +################# + +TBD \ No newline at end of file diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst new file mode 100644 index 0000000000..cf18108019 --- /dev/null +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -0,0 +1,183 @@ +.. _algo_hrnn_rnn_api_compare: + +##################### +单双层RNN API对比介绍 +##################### + +这篇教程主要介绍了 :ref:`glossary_双层RNN` 的API接口。本文中的以 :ref:`glossary_paddle` 的 :ref:`glossary_双层RNN` 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用 :ref:`glossary_双层RNN` 。本文中所有的例子，都只是介绍 :ref:`glossary_双层RNN` 的API接口，并不是使用 :ref:`glossary_双层RNN` 解决实际的问题。如果想要了解 :ref:`glossary_双层RNN` 在具体问题中的使用，请参考 :ref:`algo_hrnn_demo` 。文章中示例所使用的单元测试文件是 `test_RecurrentGradientMachine.cpp `_ 。 + +示例1：双层RNN，子序列间无Memory +================================ + + + +配置：单层RNN（:code:`sequence_layer_group`）和双层RNN（:code:`sequence_nest_layer_group`），语义完全相同。 + +读取双层序列的方法 +------------------ + +首先，我们看一下单双层序列的不同数据组织形式（您也可以采用别的组织形式）\: + +- 单层序列的数据（ :code:`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 + +.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg + :language: text + + +- 双层序列的数据（ :code:`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 + +.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest + :language: text + +其次，我们看一下单双层序列的不同dataprovider（见 :code:`sequenceGen.py` ）： + +- 单层序列的dataprovider如下： + + - word_slot是integer_value_sequence类型，代表单层序列。 + - label是integer_value类型，代表一个向量。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py + :language: python + :lines: 21-39 + +- 双层序列的dataprovider如下： + + - word_slot是integer_value_sub_sequence类型，代表双层序列。 + - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 + - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py + :language: python + :lines: 42-71 + +模型中的配置 +------------ + +首先，我们看一下单层序列的配置（见 :code:`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf + :language: python + :lines: 38-63 + + +其次，我们看一下语义相同的双层序列配置（见 :code:`sequence_nest_layer_group.conf` ），并对其详细分析： + +- batchsize=2表示一次过2句双层序列。但从上面的数据格式可知，2句双层序列和5句单层序列的数据完全一样。 +- data_layer和embedding_layer不关心数据是否是序列格式，因此两个配置在这两层上的输出是一样的。 +- lstmemory\: + + - 单层序列过了一个mixed_layer和lstmemory_group。 + - 双层序列在同样的mixed_layer和lstmemory_group外，直接加了一层group。由于这个外层group里面没有memory，表示subseq间不存在联系，即起到的作用仅仅是把双层seq拆成单层，因此双层序列过完lstmemory的输出和单层的一样。 + +- last_seq\: + + - 单层序列直接取了最后一个元素 + - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 + - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf + :language: python + :lines: 38-84 + +示例2：双进双出，subseq间有memory +================================= + +配置：单层RNN（ :code:`sequence_rnn.conf` ），双层RNN（ :code:`sequence_nest_rnn.conf` 和 :code:`sequence_nest_rnn_readonly_memory.conf` ），语义完全相同。 + +读取双层序列的方法 +------------------ + +我们看一下单双层序列的不同数据组织形式和dataprovider（见 :code:`rnn_data_provider.py`） + +.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py + :language: python + :lines: 20-32 + +- 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 +- 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 +- 单双层序列的label都分别是0和1 + +模型中的配置 +------------ + +我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 + +- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf + :language: python + :lines: 36-48 + +- 双层序列，外层memory是一个元素： + + - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 + - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf + :language: python + :lines: 39-66 + +- 双层序列，外层memory是单层序列： + + - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 + - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 + +示例3：双进双出，输入不等长 +=========================== + +.. role:: red + +.. raw:: html + + + +**输入不等长** 是指recurrent_group的多个输入在各时刻的长度可以不相等, 但需要指定一个和输出长度一致的input，用 :red:`targetInlink` 表示。参考配置：单层RNN（:code:`sequence_rnn_multi_unequalength_inputs.conf`），双层RNN（:code:`sequence_nest_rnn_multi_unequalength_inputs.conf`） + +读取双层序列的方法 +------------------ + +我们看一下单双层序列的数据组织形式和dataprovider（见 :code:`rnn_data_provider.py` ） + +.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py + :language: python + :lines: 69-97 + +data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 + +- 单层序列：两个样本分别为[[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]] 和 [[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]] +- 双层序列：两个样本分别为 + + - **样本1**\：[[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]]]。fea1和fea2都分别有2个子句，fea1=[[1, 2], [4, 5, 2]], fea2=[[5, 4, 1], [3, 1]] + - **样本2**\：[[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]]。fea1和fea2都分别有3个子句， fea1=[[0, 2], [2, 5], [0, 1, 2]], fea2=[[1, 5], [4], [2, 3, 6, 1]]。
+ - **注意**\：每个样本中，各特征的子句数目需要相等。这里说的“双进双出，输入不等长”是指fea1在i时刻的输入的长度可以不等于fea2在i时刻的输入的长度。如对于第1个样本，时刻i=2, fea1[2]=[4, 5, 2]，fea2[2]=[3, 1]，3≠2。 + +- 单双层序列中，两个样本的label都分别是0和1 + +模型中的配置 +------------ + +单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`v.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 + +- 单层序列\： + + - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 + - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf + :language: python + :lines: 41-58 + +- 双层序列\： + + - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 + - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 + - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 + +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf + :language: python + :lines: 41-89 + +示例4：beam_search的生成 +======================== + +TBD diff --git a/doc_cn/concepts/glossary.rst b/doc_cn/concepts/glossary.rst new file mode 100644 index 0000000000..a94aa73675 --- /dev/null +++ b/doc_cn/concepts/glossary.rst @@ -0,0 +1,59 @@ +.. _glossary: + +######################## +Paddle文档中使用的词汇表 +######################## + +.. _glossary_paddle: + +PaddlePaddle +------------ + +TBD + + +.. _glossary_memory: + +Memory +------ + +Memory是 :ref:`glossary_paddle` 实现 :ref:`glossary_RNN` 时候使用的一个概念。 :ref:`glossary_RNN` 即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。 + +.. graphviz:: glossary_rnn.dot + +上图中虚线的连接，即是跨越时间步的网络连接。:ref:`glossary_paddle` 在实现 :ref:`glossary_RNN` 的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的 :ref:`glossary_RNN` 实现便如下图所示。 + +.. graphviz:: glossary_rnn_with_memory.dot + +使用这种方式，:ref:`glossary_paddle` 可以比较简单的判断哪些输出是应该跨越时间步的，哪些不是。 + +.. _glossary_Sequence: + +时间序列 +-------- + +时间序列(time series)是指一系列的特征数据。这些特征数据之间的顺序是有意义的。即特征的数组，而不是特征的集合。而这每一个数组元素，或者每一个系列里的特征数据，即为一个时间步(time step)。值得注意的是，时间序列、时间步的概念，并不真正的和『时间』有关。只要一系列特征数据中的『顺序』是有意义的，即为时间序列的输入。 + +举例说明，例如文本分类中，我们通常将一句话理解成一个时间序列。比如一句话中的每一个单词，会变成词表中的位置。而这一句话就可以表示成这些位置的数组。例如 :code:`[9, 2, 3, 5, 3]` 。 + +关于时间序列(time series)的更详细准确的定义，可以参考 `维基百科页面 Time series `_ 或者 `维基百科中文页面 时间序列 `_ 。 + +另外，Paddle中经常会将时间序列成为 :code:`Sequence` 。他们在Paddle的文档和API中是一个概念。 + +.. _glossary_RNN: + +RNN +--- + +RNN 在 :ref:`glossary_paddle` 的文档中，一般表示 :code:`Recurrent neural network`，即时间递归神经网络。详细介绍可以参考 `维基百科页面 Recurrent neural network `_ 或者 `中文维基百科页面 `_ 中关于时间递归神经网络的介绍。 + +RNN 一般在 :ref:`glossary_paddle` 中，指对于一个 :ref:`glossary_Sequence` 输入数据，每一个时间步之间的神经网络具有一定的相关性。例如，某一个神经元的一个输入为上一个时间步网络中某一个神经元的输出。或者，从每一个时间步来看，神经网络的网络结构中具有有向环结构。 + +.. _glossary_双层RNN: + +双层RNN +------- + +双层RNN顾名思义，即 :ref:`glossary_RNN` 之间有一次嵌套关系。输入数据整体上是一个时间序列，而对于每一个内层特征数据而言，也是一个时间序列。即二维数组，或者数组的数组这个概念。 而双层RNN是可以处理这种输入数据的网络结构。 + +例如，对于段落的文本分类，即将一段话进行分类。我们将一段话看成句子的数组，每个句子又是单词的数组。这便是一种双层RNN的输入数据。而将这个段落的每一句话用lstm编码成一个向量，再对每一句话的编码向量用lstm编码成一个段落的向量。再对这个段落向量进行分类，即为这个双层RNN的网络结构。 diff --git a/doc_cn/concepts/glossary_rnn.dot b/doc_cn/concepts/glossary_rnn.dot new file mode 100644 index 0000000000..2cd0fb1820 --- /dev/null +++ b/doc_cn/concepts/glossary_rnn.dot @@ -0,0 +1,42 @@ +digraph G{ + subgraph cluster_timestep0 { + label="recurrent timestep i-1" + bgcolor=lightgray + node [style=filled,color=white] + fc0_0 [label="fc 0"] + fc0_1 [label="fc 1"] + fc0_2 [label="fc 2"] + + fc0_0 -> fc0_1 + fc0_1 -> fc0_2 + } + + subgraph cluster_timestep1 { + label="recurrent timestep i" + node [style=filled]; + fc1_0 [label="fc 0"] + fc1_1 [label="fc 1"] + fc1_2 [label="fc 2"] + color=blue + + fc1_0 -> fc1_1 + fc1_1 -> fc1_2 + } + + subgraph cluster_timestep2 { + label="recurrent timestep i+1" + bgcolor=lightgray + node [style=filled,color=white] + fc2_0 [label="fc 0"] + fc2_1 [label="fc 1"] + fc2_2 [label="fc 2"] + + fc2_0 -> fc2_1 + fc2_1 -> fc2_2 + } + + + fc0_1 -> fc1_1 [style="dotted" constraint=false] + fc1_1 -> fc2_1 [style="dotted" constraint=false] + +} \ No newline at end of file diff --git a/doc_cn/concepts/glossary_rnn_with_memory.dot b/doc_cn/concepts/glossary_rnn_with_memory.dot new file mode 100644 index 0000000000..0f101ec2d8 --- /dev/null +++ b/doc_cn/concepts/glossary_rnn_with_memory.dot @@ -0,0 +1,48 @@ +digraph G{ + subgraph cluster_timestep0 { + label="recurrent timestep i-1" + bgcolor=lightgray + node [style=filled,color=white] + fc0_0 [label="fc 0"] + fc0_1 [label="fc 1"] + fc0_2 [label="fc 2"] + m0 [label="memory"] + fc0_0 -> fc0_1 + fc0_1 -> fc0_2 + fc0_1 -> m0 + m0 -> fc0_1 + } + + subgraph cluster_timestep1 { + label="recurrent timestep i" + node [style=filled]; + fc1_0 [label="fc 0"] + fc1_1 [label="fc 1"] + fc1_2 [label="fc 2"] + m1 [label="memory"] + color=blue + fc1_0 -> fc1_1 + fc1_1 -> fc1_2 + fc1_1 -> m1 + m1 -> fc1_1 + } + + subgraph cluster_timestep2 { + label="recurrent timestep i+1" + bgcolor=lightgray + node [style=filled,color=white] + fc2_0 [label="fc 0"] + fc2_1 [label="fc 1"] + fc2_2 [label="fc 2"] + m2 [label="memory"] + fc2_0 -> fc2_1 + fc2_1 -> fc2_2 + fc2_1 -> m2 + m2 -> fc2_1 + } + + + m0 -> m1 [style="dotted" constraint=false] + m1 -> m2 [style="dotted" constraint=false] + +} \ No newline at end of file diff --git a/doc_cn/index.rst b/doc_cn/index.rst index f1398206fd..fef39aa527 100644 --- a/doc_cn/index.rst +++ b/doc_cn/index.rst @@ -11,6 +11,7 @@ PaddlePaddle文档 * `使用示例 `_ * `模型配置 <../doc/ui/api/trainer_config_helpers/index.html>`_ * `集群训练 `_ +* :ref:`glossary` 开发指南 -------- @@ -22,7 +23,7 @@ PaddlePaddle文档 * `Recurrent Group教程 `_ * `单层RNN示例 <../doc/algorithm/rnn/rnn.html>`_ -* `双层RNN示例 `_ +* :ref:`algo_hrnn_rnn_api_compare` * `支持双层序列作为输入的Layer `_ 常见问题 From d8cca855e1f1b69fcd9ca53ae8bd9bc6260dcfe1 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Fri, 18 Nov 2016 18:52:42 +0800 Subject: [PATCH 04/60] Refine use_concepts.rst --- doc_cn/concepts/use_concepts.rst | 97 ++++++++++++++------------------ 1 file changed, 42 insertions(+), 55 deletions(-) diff --git a/doc_cn/concepts/use_concepts.rst b/doc_cn/concepts/use_concepts.rst index 67e98edabc..d3da9cc16b 100644 --- a/doc_cn/concepts/use_concepts.rst +++ b/doc_cn/concepts/use_concepts.rst @@ -2,16 +2,16 @@ PaddlePaddle 基本使用概念 ######################### -PaddlePaddle是一个神经网络学习框架。其单机进程为 :code:`paddle train`。 单机的所有设备使用，均在单机进程内调度完成。 而多机辅助进程 :code:`paddle pserver` 负责联合多个单机进程进行通信，进而充分利用集群的计算资源。 PaddlePaddle同时以 :code:`swig api` 的形式，提供训练结果模型预测的方法和自定义训练流程。 +PaddlePaddle是一个深度学习框架，同时支持单机和多机模式的系统。命令 ``paddle train`` 可启动单机模式的进程，我们称之为 ``trainer`` 进程。单机所有设备使用均在单机进程内调度完成。多机模式除了需要启动trainer进程外，还需要通过命令 ``paddle pserver`` 启动多机参数服务器进程, 我们称之为   ``pserver`` 进程。该进程负责多个单机进程间的通信，进而充分利用集群的计算资源。 PaddlePaddle同时以 ``swig api`` 的形式，提供训练结果模型预测的方法和自定义训练流程。 -下面我们会分别介绍主要进程 :code:`paddle train` 中的一些概念。这些概念会对如何使用PaddlePaddle有一定的帮助。 了解这些概念的前提是，读者已经了解 `基本的神经网络/机器学习原理和概念 `_ 。同时，如果想要了解PaddlePaddle实现中的一些概念，请参考 `PaddlePaddle 编程中的基本概念 `_ 。 +下面我们会介绍trainer进程中的一些概念，这些概念会对如何使用PaddlePaddle有一定的帮助。 了解这些概念的前提是，读者已经了解 `基本的神经网络/机器学习原理和概念 `_ 。同时，如果想要了解PaddlePaddle实现中的一些概念，请参考 `PaddlePaddle 编程中的基本概念 `_ 。 .. contents:: -PaddlePaddle 的进程模型 -======================= +系统模块 +======== -PaddlePaddle进程内嵌了一个 :code:`python` 解释器。 这个 :code:`python` 解释器负责解析用户定义的神经网络配置，和解析用户数据，并将用户数据传入给 PaddlePaddle。 +``trainer`` 进程内嵌了一个 ``python`` 解释器， 这个 ``python`` 解释器负责解析用户定义的神经网络配置；解析输入数据流，并将数据传入给 ``trainer`` 系统。 .. graphviz:: @@ -30,95 +30,84 @@ PaddlePaddle进程内嵌了一个 :code:`python` 解释器。 这个 :code:`pyth py -> data_provider [dir="back"]; } -所以，PaddlePaddle单机训练进程，:code:`paddle train` , 对于用户的主要接口语言为 python。 主要需要用户配置的两个文件为 :code:`DataProvider` 和训练文件 :code:`TrainerConfig` 。 +所以，单机训练 ``trainer`` 进程对用户的主要接口语言为Python。用户需要配置文件主要有两个：数据流提供器 ``DataProvider`` 和模型配置 ``TrainerConfig`` 。 DataProvider ============ -DataProvider是 :code:`paddle train` 的数据提供器。 它负责将用户的原始数据转换成 PaddlePaddle 可以识别的数据类型。每当 PaddlePaddle 需要新的数据训练时，都会调用 DataProvider 返回数据。 当所有数据读取完一轮后，DataProvider 便返回空数据通知 PaddlePaddle。PaddlePaddle负责在下一轮训练开始前，将DataProvider重置。 +DataProvider是 ``trainer`` 进程的数据提供器。主要负责将用户的原始数据转换成 ``trainer`` 系统可以识别的数据类型。当系统需要新的数据训练时，会调用DataProvider获取数据接口。当所有数据读取完一轮后，DataProvider返回空数据通知系统一轮数据读取结束。 ``trainer`` 在每一轮训练开始时会重置DataProvider。 -需要注意的是，DataProvider在PaddlePaddle中是被训练逻辑调用的关系， 而不是新的数据驱动训练。并且所有的 :code:`shuffle` , 和一些随机化的噪声添加，都应该在 DataProvider 阶段完成。 +需要注意的是，DataProvider是被 ``trainer`` 系统调用，而不是新数据驱动系统；数据 ``shuffle`` 和一些随机化噪声添加都应该在DataProvider中完成。 -为了方便用户使用自己的数据格式， PaddlePaddle 提供了 `PyDataProvider`_ 来处理数据。 并且在这个Provider中，PaddlePaddle的 C++ 部分接管了如何shuffle，处理 batch，GPU/CPU通信，双缓冲，异步读取等问题。 用户可以参考 `PyDataProvider`_ 的相关文档，继续深入了解 DataProvider 的使用。 +为了用户能够灵活的处理数据，PaddlePaddle提供了处理数据的Python接口（称为 `PyDataProvider`_ ）。 在 ``PyDataProvider`` 中，系统C++模块接管了shuffle、处理batch、GPU和CPU通信、双缓冲、异步读取等问题，需要说明的是，一些情况下需要Python接口里处理shuffle，可以参考 `PyDataProvider`_ 的相关文档继续深入了解。 -训练文件 -======== +TrainerConfig +============= -训练文件是PaddlePaddle中配置神经网络结构、学习优化算法、数据传入方式的地方。 训练文件是一个python文件，使用命令行参数 :code:`--config` 传给 paddle 的主程序。 例如\: +模型配置是一个Python文件，主要包括神经网络结构、优化算法、数据传入方式，使用命令行参数 ``--config`` 传给``trainer``主程序。 例如\: .. code-block:: bash paddle train --config=trainer_config.py -一个典型简单的训练文件可能为 +一个简单的模型配置文件为： .. literalinclude:: trainer_config.py :linenos: -下面我们详细的介绍一下训练文件中各个模块的概念。 +下面我们详细的介绍一下模型配置中各个模块的概念。 trainer_config_helpers ---------------------- -PaddlePaddle的配置文件与PaddlePaddle C++端通信的最基础协议是 :code:`protobuf` 。而为了避免用户直接写比较难写的 protobuf string，我们书写了一个helpers来生成这个protobuf包。所以在文件的开始，import这些helpers函数。 +PaddlePaddle配置文件与C++模块通信的最基础协议是 ``protobuf`` 。为了避免用户直接写比较难写的protobuf string，我们通过Python代码来生成protobuf包，这就是helpers的作用。所以在文件的开始，需要import这些helpers函数。 -需要注意的是，这个 :code:`paddle.trainer_config_helpers` 包是标准的python包，这意味着用户可以选择自己喜欢的 :code:`ide` 或者编辑器来编写Paddle的配置文件，这个python包注释文档比较完善，并且考虑了IDE的代码提示与类型注释。 +需要注意的是，这个 ``paddle.trainer_config_helpers`` 包是标准的python包，这意味着用户可以选择自己喜欢的 ``IDE`` 或者编辑器来编写Paddle的配置文件，这个Python包注释文档比较完善，并提供了IDE的代码提示与类型注释。 data_sources ------------ -data_sources是配置神经网络的数据源。这里使用的函数是 :code:`define_py_data_sources2` ，这个函数是定义了使用 `PyDataProvider`_ 作为数据源。 而后缀 :code:`2` 是Paddle历史遗留问题，因为Paddle之前使用的 PyDataProvider 性能较差，所以完全重构了一个新的 `PyDataProvider`_ 。 - -data_sources里面的 train_list 和 test_list 指定的是训练文件列表和测试文件列表。 如果传入一个字符串的话，是指一个训练列表文件。这个训练列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个 list 文件，再传入给 train.list 或者 test.list 。 +data_sources配置神经网络的数据源，使用的函数是 ``define_py_data_sources2`` ，这个函数是定义了使用 `PyDataProvider`_ 提供数据源。后缀 ``2`` 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 -而 :code:`module` 和 :code:`obj` 指定了 DataProvider 的模块名和函数名。 +data_sources里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 -更具体的使用，请参考 `PyDataProvider`_ 。 +其中``module`` 和``obj``指定了DataProvider的文件名和返回数据的函数名。更详细的使用，请参考 `PyDataProvider`_ 。 settings -------- -`settings`_ 是神经网络训练算法相关的设置项。包括学习率，batch_size，优化算法，正则方法等等。具体的使用方法请参考 `settings`_ 文档。 +`settings`_ 设置训练神经网络所使用的算法。包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 网络配置 -------- -上述网络配置中余下的部分均是神经网络配置。第一行是定义一个名字叫 "pixel" 的 :code:`data_layer` 。每一个layer返回的都是一个 :code:`LayerOutput` 对象。 这里第一层的输出对象是 :code:`img` 。然后这个对象传输给了另一个 layer 函数， -:code:`simple_img_conv_pool` 。:code:`simple_img_conv_pool` 是一个组合层， -包括了图像的卷积 (convolution) 和池化(pooling)， -并继续接了一个全连接层( :code:`fc_layer` )，然后再接了一个Softmax的全连接层。 +上述配置中余下的部分是神经网络配置，主要包括网络连接、 ``cost`` 层、评估器。 -最终，网络配置输出了 :code:`classification_cost` 。标记网络输出的函数为 -:code:`outputs` 。网络的输出是神经网络的优化目标，神经网络训练的时候，实际上就是 -要最小化这个输出。 +- 首先，定义了一个名字叫"pixel"的 ``data_layer`` ，每个layer返回的都是一个 ``LayerOutput`` 对象，比如第一层的输出对象称作 ``img`` 。 +- 然后，这个对象作为另一个layer（ ``simple_img_conv_pool`` ）的输入， ``simple_img_conv_pool`` 是一个组合层，包括了图像的卷积 (convolution) 和池化(pooling)， +- 其次，连接到全连接层(``fc_layer``)，再连接到一个含Softmax激活的全连接层。 +- 最终，连接到cost层（ ``classification_cost`` ）， ``classification_cost`` 默认使用多类交叉熵损失函数和分类错误率统计评估器。标记网络输出的函数为 ``outputs`` ，网络的输出是神经网络的优化目标，神经网络训练的时候，实际上就是要最小化这个输出。 -在神经网络进行预测的时候，实际上网络的输出也是通过 :code:`outputs` 标记。 +用该模型配置进行预测时，网络的输出也是通过 ``outputs`` 标记。 Layer、Projection、Operator =========================== -PaddlePaddle的网络基本上是基于Layer来配置的。所谓的Layer即是神经网络的某一层， -而神经网络的某一层，一般是封装了许多复杂操作的操作集合。比如最简单的 -:code:`fc_layer` ，也包括矩阵乘法，多输入的求和，和activation。 +PaddlePaddle的网络是基于Layer来配置的。所谓的Layer即是神经网络的某一层，一般是封装了许多复杂操作的操作集合。比如最简单的 ``fc_layer`` ，包括矩阵乘法、多输入的求和、加Bias操作、激活( ``activation`` )函数操作。 .. code-block:: python data = data_layer(name='data', size=200) out = fc_layer(input=data, size=200, act=TanhActivation()) -而对于更灵活配置需求，可能这样基于Layer的配置是不灵活的。于是 PaddlePaddle 提供 -了基于 Projection 或者 Operator 的配置。使用Projection和Operator需要与 -:code:`mixed_layer` 配合使用。 :code:`mixed_layer` 是将layer中的元素累加求和， -并且做一个 :code:`activation` ， 而这个layer具体如何计算，是交由内部的Projection -和 Operator 定义。Projection是指含有可学习参数的操作，而Operator不含有可学习的 -参数，输入全是其他Layer的输出。 +对于更灵活配置需求，PaddlePaddle提供了基于 ``Projection`` 或者 ``Operator`` 的配置，这些需要与 ``mixed_layer`` 配合使用。 ``mixed_layer`` 是将多个输入累加求和，然后加Bias和 ``activation`` 操作。 ``mixed_layer`` 具体计算是通过内部的Projection和Operator完成。Projection含有可学习参数；而Operator不含可学习的参数，输入全是其他Layer的输出。 -例如，和 :code:`fc_layer` 同样功能的 :code:`mixed_layer` 。 +例如，和 ``fc_layer`` 同样功能的 ``mixed_layer`` 是: .. code-block:: python @@ -126,14 +115,12 @@ PaddlePaddle的网络基本上是基于Layer来配置的。所谓的Layer即是 with mixed_layer(size=200) as out: out += full_matrix_projection(input=data) -PaddlePaddle可以使用的mixed layer 配置出非常复杂的网络，甚至可以直接配置一个完整的LSTM。 -用户可以参考 `mixed_layer`_ 的相关文档进行配置。 +PaddlePaddle可以使用 ``mixed layer`` 配置出非常复杂的网络，甚至可以直接配置一个完整的LSTM。用户可以参考 `mixed_layer`_ 的相关文档进行配置。 如何利用单机的所有GPU或所有CPU核心 -================================== +=============================== -PaddlePaddle的单机进程 :code:`paddle train` 可以充分利用一台计算机上所有的GPU资 -源或者CPU。 +PaddlePaddle的单机 ``trainer`` 进程可以充分利用一台计算机上所有的GPU资源或者CPU。 如果要使用机器上多块GPU，使用如下命令即可\: @@ -145,41 +132,41 @@ PaddlePaddle的单机进程 :code:`paddle train` 可以充分利用一台计算 .. code-block:: bash - paddle train --trainer_config=4 # use 4 cpu cores. + paddle train --trainer_count=4 # use 4 cpu cores. -对于其他设置GPU的选择情况，例如选择第0、2号GPU显卡，则可以使用 :code:`CUDA_VISIBLE_DEVICES` 环境变量来选择部分的显卡。 具体可以参考连接`masking-gpus`_ 。 可以使用的命令为 +如果要指定GPU编号，例如选择第0、2号GPU，则可以设置 ``CUDA_VISIBLE_DEVICES`` 环境变量来指定特定的GPU。具体可以参考连接`masking-gpu`_ ，命令为： .. code-block:: bash - env CUDA_VISIBLE_DEVICES=0,2 paddle train --use_gpu=true --trainer_config=2 + env CUDA_VISIBLE_DEVICES=0,2 paddle train --use_gpu=true --trainer_count=2 如何利用多台机器的计算资源训练神经网络 -====================================== +=================================== -PaddlePaddle多机使用的经典方法是通过 :code:`Parameter Server` 来对多机的 :code:`paddle train` 进行同步。 而多机训练神经网络，首先要讲数据切分到不同的机器上。 切分数据文件的方式在PaddlePaddle的开源实现中并没有提供工具包。 但是切分数据并不是一件非常复杂的事情，也不是神经网络实现的重点。 +PaddlePaddle多机采用经典的 ``Parameter Server`` 架构对多个节点的 ``trainer`` 进行同步。多机训练神经网络，要讲数据切分到不同的机器上，切分数据相对简单，所以在PaddlePaddle的开源实现中并没有提供相关工具包。 -多机训练过程中，经典的拓扑结构如下\: +多机训练的经典拓扑结构如下\: .. graphviz:: pserver_topology.dot -图中每个灰色方块是一台机器，在每个机器中，先去启动一个 :code:`paddle pserver` 进程，并确定整体的端口号。可能的参数是\: +图中每个灰色方块是一台机器，在每个机器中，先启动一个 ``paddle pserver`` 进程，并指定端口号，可能的参数是\: .. code-block:: bash paddle pserver --port=5000 --num_gradient_servers=4 --nics='eth0' -这里说明系统的 :code:`paddle pserver` 的起始端口是 :code:`5000` ，并且有四个训练进程(:code:`gradient_servers`，Paddle同时将 :code:`paddle train` 进程称作 :code:`GradientServer` 。因为其为负责提供Gradient的进程)。 而对于训练进程的话，则需要在 :code:`paddle pserver` 启动之后，再在各个节点上运行如下命令\: +这里说明系统的 ``pserver`` 进程端口是 ``5000`` ，有四个训练进程(即 ``--gradient_servers=4`` ，PaddlePaddle同时将 ``trainer`` 称作 ``GradientServer`` 。因为其为负责提供Gradient)。 启动之后 ``pserver`` 进程之后，需要 ``trainer`` 训练进程，再在各个机器上运行如下命令\: .. code-block:: bash paddle train --port=5000 --pservers=192.168.100.101,192.168.100.102,192.168.100.103,192.168.100.104 --config=... -对于简单的多机协同使用上述方式即可。同时，pserver/train 通常在高级情况下，还有两个参数需要设置，他们是 +对于简单的多机协同训练使用上述方式即可。另外，pserver/train 通常在高级情况下，还需要设置下面两个参数\： * --ports_num\: 一个 pserver进程共绑定多少个端口用来做稠密更新。默认是1 * --ports_num_for_sparse\: 一个pserver进程共绑定多少端口用来做稀疏更新，默认是0 -使用手工指定端口数量，是因为Paddle的网络通信中，使用了 :code:`int32` 作为消息长度，比较容易在大模型下溢出。所以，在 :code:`paddle pserver` 进程中可以启动多个子线程去接受 trainer 的数据，这样单个子线程的长度就不会溢出了。但是这个值不可以调的过大，因为增加这个值，还是对性能，尤其是内存占用有一定的开销的，另外稀疏更新的端口如果太大的话，很容易某一个参数服务器没有分配到任何参数。 +使用手工指定端口数量，是因为Paddle的网络通信中，使用了 ``int32`` 作为消息长度，比较容易在大模型下溢出。所以，在 ``pserver`` 进程中可以启动多个子线程去接受trainer的数据，这样单个子线程的长度就不会溢出了。但是这个值不可以调的过大，因为增加这个值，对性能尤其是内存占用有一定的开销，另外稀疏更新的端口如果太大的话，很容易导致某一个参数服务器没有分配到任何参数。 详细的说明可以参考，使用 `集群训练Paddle`_ 。 From a49d1d9529e14638f84c22289b3eff8caf45da80 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Sat, 19 Nov 2016 21:37:58 +0800 Subject: [PATCH 05/60] Refine the original hrnn documentation. --- doc_cn/algorithm/rnn/hierarchical-rnn.rst | 179 ------------------ doc_cn/algorithm/rnn/hrnn_demo.rst | 2 +- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 98 +++++----- doc_cn/concepts/glossary.rst | 34 ++++ doc_cn/concepts/use_concepts.rst | 2 + doc_cn/conf.py.in | 2 +- paddle/gserver/tests/sequenceGen.py | 20 +- paddle/gserver/tests/sequence_nest_rnn.conf | 5 +- 8 files changed, 98 insertions(+), 244 deletions(-) delete mode 100644 doc_cn/algorithm/rnn/hierarchical-rnn.rst diff --git a/doc_cn/algorithm/rnn/hierarchical-rnn.rst b/doc_cn/algorithm/rnn/hierarchical-rnn.rst deleted file mode 100644 index 7c81ce8c67..0000000000 --- a/doc_cn/algorithm/rnn/hierarchical-rnn.rst +++ /dev/null @@ -1,179 +0,0 @@ -################# -双层RNN配置与示例 -################# - -我们在 :code:`paddle/gserver/tests/test_RecurrentGradientMachine` 单测中，通过多组语义相同的单双层RNN配置，讲解如何使用双层RNN。 - -示例1：双进双出，subseq间无memory -================================= - -配置：单层RNN（:code:`sequence_layer_group`）和双层RNN（:code:`sequence_nest_layer_group`），语义完全相同。 - -读取双层序列的方法 ------------------- - -首先，我们看一下单双层序列的不同数据组织形式（您也可以采用别的组织形式）\: - -- 单层序列的数据（ :code:`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 - -.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg - :language: text - - -- 双层序列的数据（ :code:`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 - -.. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest - :language: text - -其次，我们看一下单双层序列的不同dataprovider（见 :code:`sequenceGen.py` ）： - -- 单层序列的dataprovider如下： - - - word_slot是integer_value_sequence类型，代表单层序列。 - - label是integer_value类型，代表一个向量。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py - :language: python - :lines: 21-39 - -- 双层序列的dataprovider如下： - - - word_slot是integer_value_sub_sequence类型，代表双层序列。 - - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 - - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py - :language: python - :lines: 42-71 - -模型中的配置 ------------- - -首先，我们看一下单层序列的配置（见 :code:`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf - :language: python - :lines: 38-63 - - -其次，我们看一下语义相同的双层序列配置（见 :code:`sequence_nest_layer_group.conf` ），并对其详细分析： - -- batchsize=2表示一次过2句双层序列。但从上面的数据格式可知，2句双层序列和5句单层序列的数据完全一样。 -- data_layer和embedding_layer不关心数据是否是序列格式，因此两个配置在这两层上的输出是一样的。 -- lstmemory\: - - - 单层序列过了一个mixed_layer和lstmemory_group。 - - 双层序列在同样的mixed_layer和lstmemory_group外，直接加了一层group。由于这个外层group里面没有memory，表示subseq间不存在联系，即起到的作用仅仅是把双层seq拆成单层，因此双层序列过完lstmemory的输出和单层的一样。 - -- last_seq\: - - - 单层序列直接取了最后一个元素 - - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 - - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf - :language: python - :lines: 38-84 - -示例2：双进双出，subseq间有memory -================================= - -配置：单层RNN（ :code:`sequence_rnn.conf` ），双层RNN（ :code:`sequence_nest_rnn.conf` 和 :code:`sequence_nest_rnn_readonly_memory.conf` ），语义完全相同。 - -读取双层序列的方法 ------------------- - -我们看一下单双层序列的不同数据组织形式和dataprovider（见 :code:`rnn_data_provider.py`） - -.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py - :language: python - :lines: 20-32 - -- 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 -- 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 -- 单双层序列的label都分别是0和1 - -模型中的配置 ------------- - -我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 - -- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf - :language: python - :lines: 36-48 - -- 双层序列，外层memory是一个元素： - - - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf - :language: python - :lines: 39-66 - -- 双层序列，外层memory是单层序列： - - - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 - - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 - -示例3：双进双出，输入不等长 -=========================== - -.. role:: red - -.. raw:: html - - - -**输入不等长** 是指recurrent_group的多个输入在各时刻的长度可以不相等, 但需要指定一个和输出长度一致的input，用 :red:`targetInlink` 表示。参考配置：单层RNN（:code:`sequence_rnn_multi_unequalength_inputs.conf`），双层RNN（:code:`sequence_nest_rnn_multi_unequalength_inputs.conf`） - -读取双层序列的方法 ------------------- - -我们看一下单双层序列的数据组织形式和dataprovider（见 :code:`rnn_data_provider.py` ） - -.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py - :language: python - :lines: 69-97 - -data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 - -- 单层序列：两个样本分别为[[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]] 和 [[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]] -- 双层序列：两个样本分别为 - - - **样本1**\：[[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]]]。fea1和fea2都分别有2个子句，fea1=[[1, 2], [4, 5, 2]], fea2=[[5, 4, 1], [3, 1]] - - **样本2**\：[[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]]。fea1和fea2都分别有3个子句， fea1=[[0, 2], [2, 5], [0, 1, 2]], fea2=[[1, 5], [4], [2, 3, 6, 1]]。
- - **注意**\：每个样本中，各特征的子句数目需要相等。这里说的“双进双出，输入不等长”是指fea1在i时刻的输入的长度可以不等于fea2在i时刻的输入的长度。如对于第1个样本，时刻i=2, fea1[2]=[4, 5, 2]，fea2[2]=[3, 1]，3≠2。 - -- 单双层序列中，两个样本的label都分别是0和1 - -模型中的配置 ------------- - -单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`v.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 - -- 单层序列\： - - - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf - :language: python - :lines: 41-58 - -- 双层序列\： - - - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 - - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - -.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf - :language: python - :lines: 41-89 - -示例4：beam_search的生成 -======================== - -TBD diff --git a/doc_cn/algorithm/rnn/hrnn_demo.rst b/doc_cn/algorithm/rnn/hrnn_demo.rst index cf38e416c0..96396ff105 100644 --- a/doc_cn/algorithm/rnn/hrnn_demo.rst +++ b/doc_cn/algorithm/rnn/hrnn_demo.rst @@ -1,4 +1,4 @@ -.. algo_hrnn_demo: +.. _algo_hrnn_demo: ################# 双层RNN的使用示例 diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index cf18108019..8ae0f85b29 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -4,101 +4,99 @@ 单双层RNN API对比介绍 ##################### -这篇教程主要介绍了 :ref:`glossary_双层RNN` 的API接口。本文中的以 :ref:`glossary_paddle` 的 :ref:`glossary_双层RNN` 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用 :ref:`glossary_双层RNN` 。本文中所有的例子，都只是介绍 :ref:`glossary_双层RNN` 的API接口，并不是使用 :ref:`glossary_双层RNN` 解决实际的问题。如果想要了解 :ref:`glossary_双层RNN` 在具体问题中的使用，请参考 :ref:`algo_hrnn_demo` 。文章中示例所使用的单元测试文件是 `test_RecurrentGradientMachine.cpp `_ 。 +这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文中的以\ :ref:`glossary_paddle`\ 的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。文章中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 示例1：双层RNN，子序列间无Memory ================================ +在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_Sequence`\ 数据，分别进行序列操作。并且内层的序列操作之间是独立没有依赖的，即不需要使用\ :ref:`glossary_Memory`\ 的。 +在本问题中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用\ :ref:`glossary_lstm`\ 作为\ :ref:`glossary_encoder`\ ，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用\ :ref:`glossary_encoder`\ 压缩，二者语意上完全一致。这组语意相同的示例配置如下 -配置：单层RNN（:code:`sequence_layer_group`）和双层RNN（:code:`sequence_nest_layer_group`），语义完全相同。 +* 单层 \:ref:`glossary_RNN`\: `sequence_layer_group.conf `_ +* :ref:`glossary_双层RNN`\: `sequence_nest_layer_group.conf `_ -读取双层序列的方法 ------------------- -首先，我们看一下单双层序列的不同数据组织形式（您也可以采用别的组织形式）\: +读取双层序列数据 +---------------- + +首先，本示例中使用的原始数据如下\: -- 单层序列的数据（ :code:`Sequence/tour_train_wdseg`）如下，一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。 +- 本里中的原始数据一共有10个\ :ref:`glossary_sample`\ 。每个\ :ref:`glossary_sample`\ 由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg :language: text -- 双层序列的数据（ :code:`Sequence/tour_train_wdseg.nest`）如下，一共有4个样本。样本间用空行分开，代表不同的双层序列，序列数据和上面的完全一样。每个样本的子句数分别为2,3,2,3。 +- 双层序列数据一共有4个\ :ref:`glossary_sample`\ 。 每个样本间用空行分开，整体数据和原始数据完全一样。而对于双层序列的\ :ref:`glossary_lstm`\ 来说，第一条数据同时\ :ref:`glossary_encode` 两条数据成两个向量。这四条数据同时处理的句子为\ :code:`[2, 3, 2, 3]`\ 。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest :language: text -其次，我们看一下单双层序列的不同dataprovider（见 :code:`sequenceGen.py` ）： - -- 单层序列的dataprovider如下： - - - word_slot是integer_value_sequence类型，代表单层序列。 - - label是integer_value类型，代表一个向量。 +其次，对于两种不同的输入数据类型，不同\ :ref:`glossary_DataProvider`\ 对比如下(`sequenceGen.py `_)\： .. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py :language: python :lines: 21-39 + :linenos: -- 双层序列的dataprovider如下： - - - word_slot是integer_value_sub_sequence类型，代表双层序列。 - - label是integer_value_sequence类型，代表单层序列，即一个子句一个label。注意：也可以为integer_value类型，代表一个向量，即一个句子一个label。通常根据任务需求进行不同设置。 - - 关于dataprovider中input_types的详细用法，参见PyDataProvider2。 +- 这是普通的单层\ :ref:`glossary_Sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： + + * :ref:`glossary_DataProvider`\ 共返回两个数据，分别是words和label。即上述代码中的第19行。 + - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层\ :ref:`glossary_Sequence`\ 。 + - label是原始数据中对于每一句话的分类标签，它是integer_value类型的。 .. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py :language: python :lines: 42-71 + :linenos: -模型中的配置 ------------- +- 这是对于同样的数据，本示例中双层\ :ref:`glossary_Sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： + + - :ref:`glossary_DataProvider`\ 共返回两组数据，分别是sentences和labels。即在双层序列的原始数据中，每一组内的所有句子和labels + - sentences是双层\ :ref:`glossary_Sequence`\ 的数据。他内部包括了每组数据中的所有句子，又使用句子中每一个单词的词表index表示每一个句子，故为双层\ :ref:`glossary_Sequence`\ 。类型为 integer_value_sub_sequence 。 + - labels是每组内每一个句子的标签，故而是一个单层\ :ref:`glossary_Sequence`\ 。 + + +:ref:`glossary_trainer_config`\ 的模型配置 +------------------------------------------ -首先，我们看一下单层序列的配置（见 :code:`sequence_layer_group.conf`）。注意：batchsize=5表示一次过5句单层序列，因此2个batch就可以完成1个pass。 +首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行即为单层RNN序列的使用代码。这里使用了\ :ref:`glossary_paddle`\ 预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个\ :ref:`glossary_lstm`\ 网络。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf :language: python :lines: 38-63 + :linenos: + :emphasize-lines: 9-15 -其次，我们看一下语义相同的双层序列配置（见 :code:`sequence_nest_layer_group.conf` ），并对其详细分析： +其次，我们看一下语义相同的\ :ref:`glossary_双层RNN`\ 的网络配置。 -- batchsize=2表示一次过2句双层序列。但从上面的数据格式可知，2句双层序列和5句单层序列的数据完全一样。 -- data_layer和embedding_layer不关心数据是否是序列格式，因此两个配置在这两层上的输出是一样的。 -- lstmemory\: +* :ref:`glossary_paddle`\ 中的许多layer并不在意输入是否是\ :ref:`glossary_Sequence`\ ，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个\ :ref:`glossary_timestep`\ 来进行的。 - - 单层序列过了一个mixed_layer和lstmemory_group。 - - 双层序列在同样的mixed_layer和lstmemory_group外，直接加了一层group。由于这个外层group里面没有memory，表示subseq间不存在联系，即起到的作用仅仅是把双层seq拆成单层，因此双层序列过完lstmemory的输出和单层的一样。 +* 在该配置中，7-26行将双层\ :ref:`glossary_Sequence`\ 数据，先变换成单层\ :ref:`glossary_Sequence`\ 数据，在对每一个单层\ :ref:`glossary_Sequence`\ 进行处理。 -- last_seq\: + * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层\ :ref:`glossary_Sequence`\ => 单层\ :ref:`glossary_Sequence`\ ，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 + + * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个\ :ref:`glossary_lstm`\ 网络。这和单层\ :ref:`glossary_RNN`\ 的配置是等价的。 + +* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要知道使用\ :ref:`glossary_lstm` :ref:`glossary_encode`\ 成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但是，和单层\ :ref:`glossary_RNN`\ 有区别的地方是，我们是对每一个子序列取最后一个元素。于是我们设置\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 - - 单层序列直接取了最后一个元素 - - 双层序列首先（last_seq层）取了每个subseq的最后一个元素，将其拼接成一个新的单层序列；接着（expand_layer层）将其扩展成一个新的双层序列，其中第i个subseq中的所有向量均为输入的单层序列中的第i个向量；最后（average_layer层）取了每个subseq的平均值。 - - 分析得出：第一个last_seq后，每个subseq的最后一个元素就等于单层序列的最后一个元素，而expand_layer和average_layer后，依然保持每个subseq最后一个元素的值不变（这两层仅是为了展示它们的用法，实际中并不需要）。因此单双层序列的输出是一样旳。 +* 至此，\ :code:`lstm_last`\ 便和单层\ :ref:`glossary_RNN`\ 的配置中的\ :code:`lstm_last`\ 具有相同的结果了。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf :language: python - :lines: 38-84 - -示例2：双进双出，subseq间有memory -================================= + :lines: 38-64 + :linenos: + :emphasize-lines: 7-26 -配置：单层RNN（ :code:`sequence_rnn.conf` ），双层RNN（ :code:`sequence_nest_rnn.conf` 和 :code:`sequence_nest_rnn_readonly_memory.conf` ），语义完全相同。 - -读取双层序列的方法 ------------------- +示例2：:ref:`glossary_双层RNN`，子序列间有\ :ref:`glossary_Memory` +================================================================== -我们看一下单双层序列的不同数据组织形式和dataprovider（见 :code:`rnn_data_provider.py`） +本示例中，意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 同时实现一个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_Sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。而对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_Sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 -.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py - :language: python - :lines: 20-32 - -- 单层序列：有两句，分别为[1,3,2,4,5,2]和[0,2,2,5,0,1,2]。 -- 双层序列：有两句，分别为[[1,3,2],[4,5,2]]（2个子句）和[[0,2],[2,5],[0,1,2]]（3个子句）。 -- 单双层序列的label都分别是0和1 - -模型中的配置 ------------- +:ref:`glossary_trainer_config`\ 的模型配置 +------------------------------------------ 我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 diff --git a/doc_cn/concepts/glossary.rst b/doc_cn/concepts/glossary.rst index a94aa73675..518712d1fe 100644 --- a/doc_cn/concepts/glossary.rst +++ b/doc_cn/concepts/glossary.rst @@ -11,6 +11,33 @@ PaddlePaddle TBD +.. _glossary_encode: + +encode +------ + +参考\ :ref:`glossary_encoder`\ 。 + +.. _glossary_encoder: + +encoder +------- + +TBD + +.. _glossary_sample: + +样本 +---- + +TBD Sample的概念 + +.. _glossary_lstm: + +LSTM +---- + +TBD .. _glossary_memory: @@ -27,6 +54,13 @@ Memory是 :ref:`glossary_paddle` 实现 :ref:`glossary_RNN` 时候使用的一 使用这种方式，:ref:`glossary_paddle` 可以比较简单的判断哪些输出是应该跨越时间步的，哪些不是。 +.. _glossary_timestep: + +时间步 +------ + +参考 :ref:`_glossary_Sequence` 。 + .. _glossary_Sequence: 时间序列 diff --git a/doc_cn/concepts/use_concepts.rst b/doc_cn/concepts/use_concepts.rst index 67e98edabc..73fa78455f 100644 --- a/doc_cn/concepts/use_concepts.rst +++ b/doc_cn/concepts/use_concepts.rst @@ -32,6 +32,7 @@ PaddlePaddle进程内嵌了一个 :code:`python` 解释器。 这个 :code:`pyth 所以，PaddlePaddle单机训练进程，:code:`paddle train` , 对于用户的主要接口语言为 python。 主要需要用户配置的两个文件为 :code:`DataProvider` 和训练文件 :code:`TrainerConfig` 。 +.. _glossary_DataProvider: DataProvider ============ @@ -42,6 +43,7 @@ DataProvider是 :code:`paddle train` 的数据提供器。 它负责将用户的 为了方便用户使用自己的数据格式， PaddlePaddle 提供了 `PyDataProvider`_ 来处理数据。 并且在这个Provider中，PaddlePaddle的 C++ 部分接管了如何shuffle，处理 batch，GPU/CPU通信，双缓冲，异步读取等问题。 用户可以参考 `PyDataProvider`_ 的相关文档，继续深入了解 DataProvider 的使用。 +.. _glossary_trainer_config: 训练文件 ======== diff --git a/doc_cn/conf.py.in b/doc_cn/conf.py.in index 93242ace40..80e5291815 100644 --- a/doc_cn/conf.py.in +++ b/doc_cn/conf.py.in @@ -69,7 +69,7 @@ master_doc = 'index' # # This is also used if you do content translation via gettext catalogs. # Usually you set "language" from the command line for these cases. -language = None +language = 'zh_CN' # There are two options for replacing |today|: either, you set today to some # non-false value, then it is used: diff --git a/paddle/gserver/tests/sequenceGen.py b/paddle/gserver/tests/sequenceGen.py index fab876fd30..99440ada53 100644 --- a/paddle/gserver/tests/sequenceGen.py +++ b/paddle/gserver/tests/sequenceGen.py @@ -33,10 +33,10 @@ def process(settings, file_name): label, comment = line.strip().split('\t') label = int(''.join(label.split())) words = comment.split() - word_slot = [ + words = [ settings.word_dict[w] for w in words if w in settings.word_dict ] - yield word_slot, label + yield words, label ## for hierarchical sequence network @@ -52,20 +52,20 @@ def hook2(settings, dict_file, **kwargs): @provider(init_hook=hook2, should_shuffle=False) def process2(settings, file_name): with open(file_name) as fdata: - label_list = [] - word_slot_list = [] + labels = [] + sentences = [] for line in fdata: if (len(line)) > 1: label, comment = line.strip().split('\t') label = int(''.join(label.split())) words = comment.split() - word_slot = [ + words = [ settings.word_dict[w] for w in words if w in settings.word_dict ] - label_list.append(label) - word_slot_list.append(word_slot) + labels.append(label) + sentences.append(words) else: - yield word_slot_list, label_list - label_list = [] - word_slot_list = [] + yield sentences, labels + labels = [] + sentences = [] diff --git a/paddle/gserver/tests/sequence_nest_rnn.conf b/paddle/gserver/tests/sequence_nest_rnn.conf index 93b08eb2f8..524760be76 100644 --- a/paddle/gserver/tests/sequence_nest_rnn.conf +++ b/paddle/gserver/tests/sequence_nest_rnn.conf @@ -55,9 +55,8 @@ def outer_step(x): input=x) last = last_seq(input=inner_rnn_output, name="outer_rnn_state") - # "return last" should also work. But currently RecurrentGradientMachine - # does not handle it, and will report error: In hierachical RNN, all out - # links should be from sequences now. + # "return last" won't work, because recurrent_group only support the input + # sequence type is same as return sequence type. return inner_rnn_output out = recurrent_group( From a146fcf8ad37c111cd0fc44378dd6ca6804b3dfe Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Sun, 20 Nov 2016 21:25:25 +0800 Subject: [PATCH 06/60] stash --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 8ae0f85b29..eea220c043 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -98,23 +98,28 @@ :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 - -- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 +本例配置了两个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层序列模型的配置如下: .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf :language: python :lines: 36-48 + :linenos: -- 双层序列，外层memory是一个元素： +在该配置中，名称为\ :code:`rnn_state`\ 的全连接层暂存到了\ :ref:`glossary_Memory`\ 中。这个\ :ref:`glossary_Memory`\ 变量\ :code:`mem`\ 中可以保存到上一个\ :ref:`glossary_timestep`\ 中的全连接层的输出。从而实现一个全连接的\ :ref:`glossary_RNN`\ 。 - - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 +而对于\ :ref:`glossary_双层RNN`\ 来说，等价的网络配置如下\: .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf :language: python :lines: 39-66 +- 双层序列，外层memory是一个元素： + + - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 + - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 + + + - 双层序列，外层memory是单层序列： - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 From 4fcf01a849743965b8fb49db9173217e4aed2dfc Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 21 Nov 2016 16:36:07 +0800 Subject: [PATCH 07/60] Refine code, found a bad design --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 20 ++++++++----- .../simple_full_hierarchical_recurrent.dot | 30 +++++++++++++++++++ .../algorithm/rnn/simple_full_recurrent.dot | 19 ++++++++++++ 3 files changed, 62 insertions(+), 7 deletions(-) create mode 100644 doc_cn/algorithm/rnn/simple_full_hierarchical_recurrent.dot create mode 100644 doc_cn/algorithm/rnn/simple_full_recurrent.dot diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index eea220c043..09172c53f7 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -107,23 +107,29 @@ 在该配置中，名称为\ :code:`rnn_state`\ 的全连接层暂存到了\ :ref:`glossary_Memory`\ 中。这个\ :ref:`glossary_Memory`\ 变量\ :code:`mem`\ 中可以保存到上一个\ :ref:`glossary_timestep`\ 中的全连接层的输出。从而实现一个全连接的\ :ref:`glossary_RNN`\ 。 +以数据\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 举例，单层\ :ref:`glossary_RNN`\ 的网络图如下\: + +.. graphviz:: simple_full_recurrent.dot + 而对于\ :ref:`glossary_双层RNN`\ 来说，等价的网络配置如下\: .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf :language: python :lines: 39-66 + :linenos: + :emphasize-lines: 4-6 -- 双层序列，外层memory是一个元素： - - - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 +- 在该配置中，外层的\ :code:`outer_mem`\ 和内层的\ :code:`inner_mem`\ 两个变量配合，实现了和单层\ :ref:`glossary_RNN`\ 等价的全连接\ :ref:`glossary_RNN`\ 。 + - 外层\ :code:`outer_step`\ 中的\ :code:`outer_mem`\ 会将神经网络中每个子序列的最后一个结果记录下来。即将第18行的\ :code:`last`\ 变量记录下来。 + - 内层\ :code:`inner_step`\ 中的\ :code:`inner_mem`\ 会将神经网络中子序列中的每一个元素的结果记录下来。即将第7行的\ :code:`out`\ 变量记录下来。 + - 内层的\ :code:`inner_mem`\ 初始值是\ :code:`outer_mem`(:code:`boot_layer`)。于是前一个子序列的最后结果，是新的子序列的初试结果。即完成了简单的全连接\ :code:`glossary_RNN`\ 。 +本例中的\ :ref:`glossary_双层RNN`\ ,以数据\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`\ 举例，配置图如下\: -- 双层序列，外层memory是单层序列： +.. graphviz:: simple_full_hierarchical_recurrent.dot - - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 - - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 +这里有一点注意事项，Paddle目前实现的\ :ref:`glossary_双层RNN`\ 不完全支持内层\ :ref:`glossary_RNN`\ 的\ :ref:`glossary_Memory`\ 引用外层\ :ref:`glossary_RNN`\ 的某一层序列输入。即\ :code:`inner_mem`的\ :code:`boot_layer`\ 需要是非序列类型的，或者可以是序列类型，但是每个时间步下，序列长度是一致的。从序列类型转换为非序列类型，可以使用\ :code:`pooling_layer`, :code:`last_seq`, :code:`first_seq`\ 等操作进行转换。 示例3：双进双出，输入不等长 =========================== diff --git a/doc_cn/algorithm/rnn/simple_full_hierarchical_recurrent.dot b/doc_cn/algorithm/rnn/simple_full_hierarchical_recurrent.dot new file mode 100644 index 0000000000..ff278a0323 --- /dev/null +++ b/doc_cn/algorithm/rnn/simple_full_hierarchical_recurrent.dot @@ -0,0 +1,30 @@ +digraph G { + rankdir=LR; + + subgraph cluster_t0 { + a [label="4"] + b [label="5"] + c [label="2"] + } + + subgraph cluster_t1 { + d [label="0"] + e [label="9"] + } + + subgraph cluster_t2 { + f [label="8"] + g [label="1"] + h [label="4"] + } + + a -> b; + b -> c; + c -> d [constraint=false]; + + d -> e; + e -> f [constraint=false]; + + f -> g; + g -> h; +} \ No newline at end of file diff --git a/doc_cn/algorithm/rnn/simple_full_recurrent.dot b/doc_cn/algorithm/rnn/simple_full_recurrent.dot new file mode 100644 index 0000000000..cee281fbac --- /dev/null +++ b/doc_cn/algorithm/rnn/simple_full_recurrent.dot @@ -0,0 +1,19 @@ +digraph G { + rankdir=LR; + a [label="4"] + b [label="5"] + c [label="2"] + d [label="0"] + e [label="9"] + f [label="8"] + g [label="1"] + h [label="4"] + + a -> b; + b -> c; + c -> d; + d -> e; + e -> f; + f -> g; + g -> h; +} \ No newline at end of file From 8c2e5b2cadf450de6f5b44ffffa42da93651c3c4 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Mon, 21 Nov 2016 20:56:00 +0800 Subject: [PATCH 08/60] Update use_concepts --- doc_cn/concepts/use_concepts.rst | 156 ++++++++++++++----------------- doc_cn/faq/index.rst | 19 +++- 2 files changed, 88 insertions(+), 87 deletions(-) diff --git a/doc_cn/concepts/use_concepts.rst b/doc_cn/concepts/use_concepts.rst index d3da9cc16b..49c45ff779 100644 --- a/doc_cn/concepts/use_concepts.rst +++ b/doc_cn/concepts/use_concepts.rst @@ -2,16 +2,20 @@ PaddlePaddle 基本使用概念 ######################### -PaddlePaddle是一个深度学习框架，同时支持单机和多机模式的系统。命令 ``paddle train`` 可启动单机模式的进程，我们称之为 ``trainer`` 进程。单机所有设备使用均在单机进程内调度完成。多机模式除了需要启动trainer进程外，还需要通过命令 ``paddle pserver`` 启动多机参数服务器进程, 我们称之为   ``pserver`` 进程。该进程负责多个单机进程间的通信，进而充分利用集群的计算资源。 PaddlePaddle同时以 ``swig api`` 的形式，提供训练结果模型预测的方法和自定义训练流程。 +PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 -下面我们会介绍trainer进程中的一些概念，这些概念会对如何使用PaddlePaddle有一定的帮助。 了解这些概念的前提是，读者已经了解 `基本的神经网络/机器学习原理和概念 `_ 。同时，如果想要了解PaddlePaddle实现中的一些概念，请参考 `PaddlePaddle 编程中的基本概念 `_ 。 +单节模式用命令 ``paddle train`` 可以启动一个trainer进程，一个单机训练作业只包括一个trainer进程，单机的所有设备使用，均在单机进程内调度完成。 + +如果数据规模比较大，希望加速训练，可以启动分布式作业。一个分布式作业里包括若干trainer进程和若干Parameter Server（或称pserver）进程。用命令 ``paddle pserver`` 可以启动 pserver 进程，pserver进程用于协调多个trainer进程之间的通信。 + +本文首先介绍trainer进程中的一些使用概念，然后介绍pserver进程中概念。 .. contents:: -系统模块 +系统框图 ======== -``trainer`` 进程内嵌了一个 ``python`` 解释器， 这个 ``python`` 解释器负责解析用户定义的神经网络配置；解析输入数据流，并将数据传入给 ``trainer`` 系统。 +下图描述了用户使用框图，PaddlePaddle里链接了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为DataProvider，通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中。下面将分别介绍这两部分。 .. graphviz:: @@ -30,132 +34,105 @@ PaddlePaddle是一个深度学习框架，同时支持单机和多机模式的 py -> data_provider [dir="back"]; } -所以，单机训练 ``trainer`` 进程对用户的主要接口语言为Python。用户需要配置文件主要有两个：数据流提供器 ``DataProvider`` 和模型配置 ``TrainerConfig`` 。 - - DataProvider ============ -DataProvider是 ``trainer`` 进程的数据提供器。主要负责将用户的原始数据转换成 ``trainer`` 系统可以识别的数据类型。当系统需要新的数据训练时，会调用DataProvider获取数据接口。当所有数据读取完一轮后，DataProvider返回空数据通知系统一轮数据读取结束。 ``trainer`` 在每一轮训练开始时会重置DataProvider。 +在不同的应用里，训练数据的格式往往各不相同。因此，为了用户能够灵活的处理数据，我们提供了Python处理数据的接口，称为 `PyDataProvider`_ 。 -需要注意的是，DataProvider是被 ``trainer`` 系统调用，而不是新数据驱动系统；数据 ``shuffle`` 和一些随机化噪声添加都应该在DataProvider中完成。 +trainer进程会调用DataProvider函数，将用户的原始数据转换成系统可以识别的数据类型。当所有数据读取完一轮后，DataProvider返回空数据，通知系统一轮数据读取结束，系统每一轮训练开始时会重置DataProvider。需要注意的是，DataProvider是被系统调用，而不是新数据驱动系统，一些随机化噪声添加都应该在DataProvider中完成。 -为了用户能够灵活的处理数据，PaddlePaddle提供了处理数据的Python接口（称为 `PyDataProvider`_ ）。 在 ``PyDataProvider`` 中，系统C++模块接管了shuffle、处理batch、GPU和CPU通信、双缓冲、异步读取等问题，需要说明的是，一些情况下需要Python接口里处理shuffle，可以参考 `PyDataProvider`_ 的相关文档继续深入了解。 +在 ``PyDataProvider`` 中，系统C++模块接管了shuffle、处理batch、GPU和CPU通信、双缓冲、异步读取等问题，一些情况下(如：``min_pool_size=0``)需要Python接口里处理shuffle，可以参考 `PyDataProvider`_ 的相关文档继续深入了解。 -TrainerConfig -============= - -模型配置是一个Python文件，主要包括神经网络结构、优化算法、数据传入方式，使用命令行参数 ``--config`` 传给``trainer``主程序。 例如\: - -.. code-block:: bash +模型配置文件 +============ - paddle train --config=trainer_config.py +模型配置主要包括数据传入接口定义(DataConfig)、优化算法(OptimizationConfig)、网络结构(ModelConfig)。 其中数据传入接口定义与DataProvider的关系是：DataProvider里定义数据读取函数，配置文件的DataConfig里指定DataProvider文件名字、生成数据函数接口，请不要混淆。 一个简单的模型配置文件为： .. literalinclude:: trainer_config.py :linenos: -下面我们详细的介绍一下模型配置中各个模块的概念。 - +文件开头 ``from paddle.trainer_config_helpers import *`` ，是因为PaddlePaddle配置文件与C++模块通信的最基础协议是protobuf。为了避免用户直接写复杂的protobuf string，我们为用户定以Python接口来配置网络，该Python代码可以生成protobuf包，这就是的作用`trainer_config_helpers`_的作用。因此，在文件的开始，需要import这些函数。 这个包里面包含了模型配置需要的各个模块。 -trainer_config_helpers ----------------------- +下面分别介绍DataConfig、OptimizationConfig、ModelConfig这三部分该概念。 -PaddlePaddle配置文件与C++模块通信的最基础协议是 ``protobuf`` 。为了避免用户直接写比较难写的protobuf string，我们通过Python代码来生成protobuf包，这就是helpers的作用。所以在文件的开始，需要import这些helpers函数。 +DataConfig +---------- -需要注意的是，这个 ``paddle.trainer_config_helpers`` 包是标准的python包，这意味着用户可以选择自己喜欢的 ``IDE`` 或者编辑器来编写Paddle的配置文件，这个Python包注释文档比较完善，并提供了IDE的代码提示与类型注释。 +使用函数 ``define_py_data_sources2`` 配置数据源，后缀 2 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 -data_sources ------------- +``define_py_data_sources2`` 里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 -data_sources配置神经网络的数据源，使用的函数是 ``define_py_data_sources2`` ，这个函数是定义了使用 `PyDataProvider`_ 提供数据源。后缀 ``2`` 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 +``module`` 和 ``obj`` 指定了DataProvider的文件名和返回数据的函数名。更详细的使用，请参考 `PyDataProvider`_ 。 -data_sources里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 +OptimizationConfig +------------------ -其中``module`` 和``obj``指定了DataProvider的文件名和返回数据的函数名。更详细的使用，请参考 `PyDataProvider`_ 。 +通过`settings`_ 接口设置神经网络所使用的训练参数和优化算法，包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 -settings --------- +ModelConfig +----------- -`settings`_ 设置训练神经网络所使用的算法。包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 +神经网络配置主要包括网络连接、激活函数、损失函数、评估器。 -网络配置 --------- +- 网络连接： 主要由Layer组成，每个Layer返回的都是一个 ``LayerOutput`` 对象，Layer里面可以定义参数属性、激活类型等。 -上述配置中余下的部分是神经网络配置，主要包括网络连接、 ``cost`` 层、评估器。 + 为了更灵活的配置，PaddlePaddle提供了基于 Projection 或者 Operator 的配置，这两个需要与 ``mixed_layer`` 配合使用。这里简单介绍Layer、Projection、Operator的概念: -- 首先，定义了一个名字叫"pixel"的 ``data_layer`` ，每个layer返回的都是一个 ``LayerOutput`` 对象，比如第一层的输出对象称作 ``img`` 。 -- 然后，这个对象作为另一个layer（ ``simple_img_conv_pool`` ）的输入， ``simple_img_conv_pool`` 是一个组合层，包括了图像的卷积 (convolution) 和池化(pooling)， -- 其次，连接到全连接层(``fc_layer``)，再连接到一个含Softmax激活的全连接层。 -- 最终，连接到cost层（ ``classification_cost`` ）， ``classification_cost`` 默认使用多类交叉熵损失函数和分类错误率统计评估器。标记网络输出的函数为 ``outputs`` ，网络的输出是神经网络的优化目标，神经网络训练的时候，实际上就是要最小化这个输出。 + - Layer: 神经网络的某一层，可以有可学习的参数，一般是封装了许多复杂操作的集合。 + - Projection：需要与 ``mixed_layer`` 配合使用，含可学习参数。 + - Operator： 需要与 ``mixed_layer`` 配合使用，不含可学习参数，输入全是其他Layer的输出。 -用该模型配置进行预测时，网络的输出也是通过 ``outputs`` 标记。 + + 这个配置文件网络由 ``data_layer`` 、 ``simple_img_conv_pool`` 、 ``fc_layer`` 组成。 + - `data_layer`_ ： 通常每个配置文件都会包括 ``data_layer`` ，定义输入数据大小。 + - `simple_img_conv_pool`_ ：是一个组合层，包括了图像的卷积 (convolution)和池化(pooling)。 + - `fc_layer`_ ：全连接层，激活函数为Softmax，这里也可叫分类层。 -Layer、Projection、Operator -=========================== + +- 损失函数和评估器：损失函数即为网络的优化目标，评估器可以评价模型结果。 -PaddlePaddle的网络是基于Layer来配置的。所谓的Layer即是神经网络的某一层，一般是封装了许多复杂操作的操作集合。比如最简单的 ``fc_layer`` ，包括矩阵乘法、多输入的求和、加Bias操作、激活( ``activation`` )函数操作。 - -.. code-block:: python + PaddlePaddle包括很多损失函数和评估起，详细可以参考 `损失函数层`_ 和 `评估器`_ 。这里 ``classification_cost`` 默认使用多类交叉熵损失函数和分类错误率统计评估器。 + +- ``outputs``: 标记网络输出的函数为 ``outputs`` 。 - data = data_layer(name='data', size=200) - out = fc_layer(input=data, size=200, act=TanhActivation()) + 训练阶段，网络的输出为神经网络的优化目标；预测阶段，网络的输出也可通过 ``outputs`` 标记。 -对于更灵活配置需求，PaddlePaddle提供了基于 ``Projection`` 或者 ``Operator`` 的配置，这些需要与 ``mixed_layer`` 配合使用。 ``mixed_layer`` 是将多个输入累加求和，然后加Bias和 ``activation`` 操作。 ``mixed_layer`` 具体计算是通过内部的Projection和Operator完成。Projection含有可学习参数；而Operator不含可学习的参数，输入全是其他Layer的输出。 +这里对 ``mixed_layer`` 稍做详细说明， 该Layer将多个输入(Projection 或 Operator)累加求和，具体计算是通过内部的 Projection 和 Operator 完成，然后加 Bias 和 activation 操作， 例如，和 ``fc_layer`` 同样功能的 ``mixed_layer`` 是: .. code-block:: python + + data = data_layer(name='data', size=200) + with mixed_layer(size=200) as out: + out += full_matrix_projection(input=data) - data = data_layer(name='data', size=200) - with mixed_layer(size=200) as out: - out += full_matrix_projection(input=data) - -PaddlePaddle可以使用 ``mixed layer`` 配置出非常复杂的网络，甚至可以直接配置一个完整的LSTM。用户可以参考 `mixed_layer`_ 的相关文档进行配置。 - -如何利用单机的所有GPU或所有CPU核心 -=============================== - -PaddlePaddle的单机 ``trainer`` 进程可以充分利用一台计算机上所有的GPU资源或者CPU。 +PaddlePaddle 可以使用 ``mixed layer`` 配置出非常复杂的网络，甚至可以直接配置一个完整的LSTM。用户可以参考 `mixed_layer`_ 的相关文档进行配置。 -如果要使用机器上多块GPU，使用如下命令即可\: -.. code-block:: bash - - paddle train --use_gpu=true --trainer_count=4 # use 4 gpu card, 0, 1, 2, 3 - -如果要使用机器上多块CPU, 使用如下命令即可\: - -.. code-block:: bash - - paddle train --trainer_count=4 # use 4 cpu cores. - -如果要指定GPU编号，例如选择第0、2号GPU，则可以设置 ``CUDA_VISIBLE_DEVICES`` 环境变量来指定特定的GPU。具体可以参考连接`masking-gpu`_ ，命令为： +分布式训练 +========== -.. code-block:: bash - - env CUDA_VISIBLE_DEVICES=0,2 paddle train --use_gpu=true --trainer_count=2 - -如何利用多台机器的计算资源训练神经网络 -=================================== - -PaddlePaddle多机采用经典的 ``Parameter Server`` 架构对多个节点的 ``trainer`` 进行同步。多机训练神经网络，要讲数据切分到不同的机器上，切分数据相对简单，所以在PaddlePaddle的开源实现中并没有提供相关工具包。 - -多机训练的经典拓扑结构如下\: +PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trainer 进行同步。多机训练的经典拓扑结构如下\: .. graphviz:: pserver_topology.dot -图中每个灰色方块是一台机器，在每个机器中，先启动一个 ``paddle pserver`` 进程，并指定端口号，可能的参数是\: +图中每个灰色方块是一台机器，在每个机器中，先使用命令 ``paddle pserver`` 启动一个pserver进程，并指定端口号，可能的参数是\: .. code-block:: bash paddle pserver --port=5000 --num_gradient_servers=4 --nics='eth0' -这里说明系统的 ``pserver`` 进程端口是 ``5000`` ，有四个训练进程(即 ``--gradient_servers=4`` ，PaddlePaddle同时将 ``trainer`` 称作 ``GradientServer`` 。因为其为负责提供Gradient)。 启动之后 ``pserver`` 进程之后，需要 ``trainer`` 训练进程，再在各个机器上运行如下命令\: +* 指定 pserver 进程端口是 5000 。 +* 有四个训练进程(即 ``--gradient_servers=4`` ，PaddlePaddle同时将 trainer 称作 GradientServer 。因为其为负责提供Gradient) 。 +* 指定以太网类型为TCP网络。 + +启动之后 pserver 进程之后，需要启动 trainer 训练进程，在各个机器上运行如下命令\: .. code-block:: bash @@ -163,16 +140,23 @@ PaddlePaddle多机采用经典的 ``Parameter Server`` 架构对多个节点的 对于简单的多机协同训练使用上述方式即可。另外，pserver/train 通常在高级情况下，还需要设置下面两个参数\： -* --ports_num\: 一个 pserver进程共绑定多少个端口用来做稠密更新。默认是1 +* --ports_num\: 一个 pserver 进程共绑定多少个端口用来做稠密更新。默认是1 * --ports_num_for_sparse\: 一个pserver进程共绑定多少端口用来做稀疏更新，默认是0 -使用手工指定端口数量，是因为Paddle的网络通信中，使用了 ``int32`` 作为消息长度，比较容易在大模型下溢出。所以，在 ``pserver`` 进程中可以启动多个子线程去接受trainer的数据，这样单个子线程的长度就不会溢出了。但是这个值不可以调的过大，因为增加这个值，对性能尤其是内存占用有一定的开销，另外稀疏更新的端口如果太大的话，很容易导致某一个参数服务器没有分配到任何参数。 +使用手工指定端口数量，是因为Paddle的网络通信中，使用了 int32 作为消息长度，比较容易在大模型下溢出。所以，在 pserver 进程中可以启动多个子线程去接受 trainer 的数据，这样单个子线程的长度就不会溢出了。但是这个值不可以调的过大，因为增加这个值，对性能尤其是内存占用有一定的开销，另外稀疏更新的端口如果太大的话，很容易导致某一个参数服务器没有分配到任何参数。 详细的说明可以参考，使用 `集群训练Paddle`_ 。 .. _PyDataProvider: ../ui/data_provider/pydataprovider2.html -.. _settings: ../../doc/ui/api/trainer_config_helpers/optimizers.html#settings -.. _mixed_layer: ../../doc/ui/api/trainer_config_helpers/layers.html#mixed-layer -.. _masking-gpu: http://www.acceleware.com/blog/cudavisibledevices-masking-gpus +.. _settings: ../../doc/ui/api/trainer_config_helpers/optimizers.html#settings +.. _trainer_config_helper: ../../doc/ui/api/trainer_config_helpers/index.html +.. _data_layer: ../../doc/ui/api/trainer_config_helpers/layers.html#data-layer +.. _simple_img_conv_pool: ../../doc/ui/api/trainer_config_helpers/networks.html#simple-img-conv-pool +.. _fc_layer: ../../doc/ui/api/trainer_config_helpers/layers.html#fc-layer +.. _损失函数层: ../../doc/ui/api/trainer_config_helpers/layers.html#cost-layers +.. _评估器: ../../doc/ui/api/trainer_config_helpers/evaluators.html +.. _mixed_layer: ../../doc/ui/api/trainer_config_helpers/layers.html#mixed-layer +.. _masking-gpu: http://www.acceleware.com/blog/cudavisibledevices-masking-gpus + .. _集群训练Paddle: ../cluster/index.html diff --git a/doc_cn/faq/index.rst b/doc_cn/faq/index.rst index 3eb0e10ae2..8da21e5b8b 100644 --- a/doc_cn/faq/index.rst +++ b/doc_cn/faq/index.rst @@ -213,4 +213,21 @@ PaddlePaddle的参数使用名字 :code:`name` 作为参数的ID，相同名字 原因是：单元测试使用了一个旧版本的python包，而没有测试到代码中实际修改的python包。即单元测试需要一个干净的环境： * 如果paddle包已经在python的site-packages里面了，那么单元测试时使用的paddle包，就是site-packages里面的python包，而不是源码目录里 :code:`/python` 目录下的python包。 -* 即便设置了 :code:`PYTHONPATH` 到 :code:`/python` 也没用，因为python的搜索路径是优先已经安装的python包。 \ No newline at end of file +* 即便设置了 :code:`PYTHONPATH` 到 :code:`/python` 也没用，因为python的搜索路径是优先已经安装的python包。 + +9. 如何指定GPU设备 +----------------- + +例如机器上有4块GPU，编号从0开始，指定使用2、3号GPU： + +* 方式1：通过 ``CUDA_VISIBLE_DEVICES`` 环境变量来指定特定的GPU。 + +.. code-block:: bash + + env CUDA_VISIBLE_DEVICES=2,3 paddle train --use_gpu=true --trainer_count=2 + +* 方式2：通过命令行参数 ``--gpu_id`` 指定。 + +.. code-block:: bash + + paddle train --use_gpu=true --trainer_count=2 --gpu_id=2 From f56643dd902168cf07c64339c3b7d8c74868a5da Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 23 Nov 2016 15:49:06 +0800 Subject: [PATCH 09/60] Remove glossary --- .../rnn}/glossary_rnn.dot | 0 .../rnn}/glossary_rnn_with_memory.dot | 0 doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 124 ++++++++++++------ doc_cn/concepts/glossary.rst | 93 ------------- doc_cn/index.rst | 1 - 5 files changed, 86 insertions(+), 132 deletions(-) rename doc_cn/{concepts => algorithm/rnn}/glossary_rnn.dot (100%) rename doc_cn/{concepts => algorithm/rnn}/glossary_rnn_with_memory.dot (100%) delete mode 100644 doc_cn/concepts/glossary.rst diff --git a/doc_cn/concepts/glossary_rnn.dot b/doc_cn/algorithm/rnn/glossary_rnn.dot similarity index 100% rename from doc_cn/concepts/glossary_rnn.dot rename to doc_cn/algorithm/rnn/glossary_rnn.dot diff --git a/doc_cn/concepts/glossary_rnn_with_memory.dot b/doc_cn/algorithm/rnn/glossary_rnn_with_memory.dot similarity index 100% rename from doc_cn/concepts/glossary_rnn_with_memory.dot rename to doc_cn/algorithm/rnn/glossary_rnn_with_memory.dot diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 09172c53f7..896dd7ada9 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -4,16 +4,16 @@ 单双层RNN API对比介绍 ##################### -这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文中的以\ :ref:`glossary_paddle`\ 的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。文章中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 +这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文中的以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。文章中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 示例1：双层RNN，子序列间无Memory ================================ -在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_Sequence`\ 数据，分别进行序列操作。并且内层的序列操作之间是独立没有依赖的，即不需要使用\ :ref:`glossary_Memory`\ 的。 +在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_sequence`\ 数据，分别进行序列操作。并且内层的序列操作之间是独立没有依赖的，即不需要使用\ :ref:`glossary_Memory`\ 的。 -在本问题中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用\ :ref:`glossary_lstm`\ 作为\ :ref:`glossary_encoder`\ ，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用\ :ref:`glossary_encoder`\ 压缩，二者语意上完全一致。这组语意相同的示例配置如下 +在本问题中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用encoder压缩，二者语意上完全一致。这组语意相同的示例配置如下 -* 单层 \:ref:`glossary_RNN`\: `sequence_layer_group.conf `_ +* 单层\ :ref:`glossary_RNN`\: `sequence_layer_group.conf `_ * :ref:`glossary_双层RNN`\: `sequence_nest_layer_group.conf `_ @@ -22,13 +22,13 @@ 首先，本示例中使用的原始数据如下\: -- 本里中的原始数据一共有10个\ :ref:`glossary_sample`\ 。每个\ :ref:`glossary_sample`\ 由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 +- 本里中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg :language: text -- 双层序列数据一共有4个\ :ref:`glossary_sample`\ 。 每个样本间用空行分开，整体数据和原始数据完全一样。而对于双层序列的\ :ref:`glossary_lstm`\ 来说，第一条数据同时\ :ref:`glossary_encode` 两条数据成两个向量。这四条数据同时处理的句子为\ :code:`[2, 3, 2, 3]`\ 。 +- 双层序列数据一共有4个样本。 每个样本间用空行分开，整体数据和原始数据完全一样。而对于双层序列的LSTM来说，第一条数据同时encode两条数据成两个向量。这四条数据同时处理的句子为\ :code:`[2, 3, 2, 3]`\ 。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest :language: text @@ -40,10 +40,10 @@ :lines: 21-39 :linenos: -- 这是普通的单层\ :ref:`glossary_Sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： +- 这是普通的单层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： * :ref:`glossary_DataProvider`\ 共返回两个数据，分别是words和label。即上述代码中的第19行。 - - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层\ :ref:`glossary_Sequence`\ 。 + - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层\ :ref:`glossary_sequence`\ 。 - label是原始数据中对于每一句话的分类标签，它是integer_value类型的。 .. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py @@ -51,17 +51,17 @@ :lines: 42-71 :linenos: -- 这是对于同样的数据，本示例中双层\ :ref:`glossary_Sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： +- 这是对于同样的数据，本示例中双层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： - :ref:`glossary_DataProvider`\ 共返回两组数据，分别是sentences和labels。即在双层序列的原始数据中，每一组内的所有句子和labels - - sentences是双层\ :ref:`glossary_Sequence`\ 的数据。他内部包括了每组数据中的所有句子，又使用句子中每一个单词的词表index表示每一个句子，故为双层\ :ref:`glossary_Sequence`\ 。类型为 integer_value_sub_sequence 。 - - labels是每组内每一个句子的标签，故而是一个单层\ :ref:`glossary_Sequence`\ 。 + - sentences是双层\ :ref:`glossary_sequence`\ 的数据。他内部包括了每组数据中的所有句子，又使用句子中每一个单词的词表index表示每一个句子，故为双层\ :ref:`glossary_sequence`\ 。类型为 integer_value_sub_sequence 。 + - labels是每组内每一个句子的标签，故而是一个单层\ :ref:`glossary_sequence`\ 。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行即为单层RNN序列的使用代码。这里使用了\ :ref:`glossary_paddle`\ 预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个\ :ref:`glossary_lstm`\ 网络。 +首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个LSTM网络。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf :language: python @@ -72,15 +72,15 @@ 其次，我们看一下语义相同的\ :ref:`glossary_双层RNN`\ 的网络配置。 -* :ref:`glossary_paddle`\ 中的许多layer并不在意输入是否是\ :ref:`glossary_Sequence`\ ，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个\ :ref:`glossary_timestep`\ 来进行的。 +* PaddlePaddle中的许多layer并不在意输入是否是\ :ref:`glossary_sequence`\ ，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个\ :ref:`glossary_timestep`\ 来进行的。 -* 在该配置中，7-26行将双层\ :ref:`glossary_Sequence`\ 数据，先变换成单层\ :ref:`glossary_Sequence`\ 数据，在对每一个单层\ :ref:`glossary_Sequence`\ 进行处理。 +* 在该配置中，7-26行将双层\ :ref:`glossary_sequence`\ 数据，先变换成单层\ :ref:`glossary_sequence`\ 数据，在对每一个单层\ :ref:`glossary_sequence`\ 进行处理。 - * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层\ :ref:`glossary_Sequence`\ => 单层\ :ref:`glossary_Sequence`\ ，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 + * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层\ :ref:`glossary_sequence`\ => 单层\ :ref:`glossary_sequence`\ ，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 - * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个\ :ref:`glossary_lstm`\ 网络。这和单层\ :ref:`glossary_RNN`\ 的配置是等价的。 + * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层\ :ref:`glossary_RNN`\ 的配置是等价的。 -* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要知道使用\ :ref:`glossary_lstm` :ref:`glossary_encode`\ 成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但是，和单层\ :ref:`glossary_RNN`\ 有区别的地方是，我们是对每一个子序列取最后一个元素。于是我们设置\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 +* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要知道使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但是，和单层\ :ref:`glossary_RNN`\ 有区别的地方是，我们是对每一个子序列取最后一个元素。于是我们设置\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 * 至此，\ :code:`lstm_last`\ 便和单层\ :ref:`glossary_RNN`\ 的配置中的\ :code:`lstm_last`\ 具有相同的结果了。 @@ -93,43 +93,32 @@ 示例2：:ref:`glossary_双层RNN`，子序列间有\ :ref:`glossary_Memory` ================================================================== -本示例中，意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 同时实现一个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_Sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。而对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_Sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 +本示例中，意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 同时实现一个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。而对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -本例配置了两个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层序列模型的配置如下: +我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 + +- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf :language: python :lines: 36-48 - :linenos: - -在该配置中，名称为\ :code:`rnn_state`\ 的全连接层暂存到了\ :ref:`glossary_Memory`\ 中。这个\ :ref:`glossary_Memory`\ 变量\ :code:`mem`\ 中可以保存到上一个\ :ref:`glossary_timestep`\ 中的全连接层的输出。从而实现一个全连接的\ :ref:`glossary_RNN`\ 。 -以数据\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 举例，单层\ :ref:`glossary_RNN`\ 的网络图如下\: +- 双层序列，外层memory是一个元素： -.. graphviz:: simple_full_recurrent.dot - -而对于\ :ref:`glossary_双层RNN`\ 来说，等价的网络配置如下\: + - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 + - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf :language: python :lines: 39-66 - :linenos: - :emphasize-lines: 4-6 -- 在该配置中，外层的\ :code:`outer_mem`\ 和内层的\ :code:`inner_mem`\ 两个变量配合，实现了和单层\ :ref:`glossary_RNN`\ 等价的全连接\ :ref:`glossary_RNN`\ 。 +- 双层序列，外层memory是单层序列： - - 外层\ :code:`outer_step`\ 中的\ :code:`outer_mem`\ 会将神经网络中每个子序列的最后一个结果记录下来。即将第18行的\ :code:`last`\ 变量记录下来。 - - 内层\ :code:`inner_step`\ 中的\ :code:`inner_mem`\ 会将神经网络中子序列中的每一个元素的结果记录下来。即将第7行的\ :code:`out`\ 变量记录下来。 - - 内层的\ :code:`inner_mem`\ 初始值是\ :code:`outer_mem`(:code:`boot_layer`)。于是前一个子序列的最后结果，是新的子序列的初试结果。即完成了简单的全连接\ :code:`glossary_RNN`\ 。 - -本例中的\ :ref:`glossary_双层RNN`\ ,以数据\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`\ 举例，配置图如下\: - -.. graphviz:: simple_full_hierarchical_recurrent.dot - -这里有一点注意事项，Paddle目前实现的\ :ref:`glossary_双层RNN`\ 不完全支持内层\ :ref:`glossary_RNN`\ 的\ :ref:`glossary_Memory`\ 引用外层\ :ref:`glossary_RNN`\ 的某一层序列输入。即\ :code:`inner_mem`的\ :code:`boot_layer`\ 需要是非序列类型的，或者可以是序列类型，但是每个时间步下，序列长度是一致的。从序列类型转换为非序列类型，可以使用\ :code:`pooling_layer`, :code:`last_seq`, :code:`first_seq`\ 等操作进行转换。 + - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 + - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 示例3：双进双出，输入不等长 =========================== @@ -190,3 +179,62 @@ data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 ======================== TBD + + +词汇表 +====== + +.. _glossary_memory: + +Memory +------ + +Memory是PaddlePaddle实现 :ref:`glossary_RNN` 时候使用的一个概念。 :ref:`glossary_RNN` 即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。 + +.. graphviz:: glossary_rnn.dot + +上图中虚线的连接，即是跨越时间步的网络连接。PaddlePaddle在实现 :ref:`glossary_RNN` 的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的 :ref:`glossary_RNN` 实现便如下图所示。 + +.. graphviz:: glossary_rnn_with_memory.dot + +使用这种方式，PaddlePaddle可以比较简单的判断哪些输出是应该跨越时间步的，哪些不是。 + +.. _glossary_timestep: + +时间步 +------ + +参考 :ref:`glossary_sequence` 。 + + +.. _glossary_sequence: + +时间序列 +-------- + +时间序列(time series)是指一系列的特征数据。这些特征数据之间的顺序是有意义的。即特征的数组，而不是特征的集合。而这每一个数组元素，或者每一个系列里的特征数据，即为一个时间步(time step)。值得注意的是，时间序列、时间步的概念，并不真正的和『时间』有关。只要一系列特征数据中的『顺序』是有意义的，即为时间序列的输入。 + +举例说明，例如文本分类中，我们通常将一句话理解成一个时间序列。比如一句话中的每一个单词，会变成词表中的位置。而这一句话就可以表示成这些位置的数组。例如 :code:`[9, 2, 3, 5, 3]` 。 + +关于时间序列(time series)的更详细准确的定义，可以参考 `维基百科页面 Time series `_ 或者 `维基百科中文页面 时间序列 `_ 。 + +另外，Paddle中经常会将时间序列成为 :code:`Sequence` 。他们在Paddle的文档和API中是一个概念。 + +.. _glossary_RNN: + +RNN +--- + +RNN 在PaddlePaddle的文档中，一般表示 :code:`Recurrent neural network`，即时间递归神经网络。详细介绍可以参考 `维基百科页面 Recurrent neural network `_ 或者 `中文维基百科页面 `_ 中关于时间递归神经网络的介绍。 + +RNN 一般在PaddlePaddle中，指对于一个 :ref:`glossary_sequence` 输入数据，每一个时间步之间的神经网络具有一定的相关性。例如，某一个神经元的一个输入为上一个时间步网络中某一个神经元的输出。或者，从每一个时间步来看，神经网络的网络结构中具有有向环结构。 + +.. _glossary_双层RNN: + +双层RNN +------- + +双层RNN顾名思义，即 :ref:`glossary_RNN` 之间有一次嵌套关系。输入数据整体上是一个时间序列，而对于每一个内层特征数据而言，也是一个时间序列。即二维数组，或者数组的数组这个概念。 而双层RNN是可以处理这种输入数据的网络结构。 + +例如，对于段落的文本分类，即将一段话进行分类。我们将一段话看成句子的数组，每个句子又是单词的数组。这便是一种双层RNN的输入数据。而将这个段落的每一句话用lstm编码成一个向量，再对每一句话的编码向量用lstm编码成一个段落的向量。再对这个段落向量进行分类，即为这个双层RNN的网络结构。 + diff --git a/doc_cn/concepts/glossary.rst b/doc_cn/concepts/glossary.rst deleted file mode 100644 index 518712d1fe..0000000000 --- a/doc_cn/concepts/glossary.rst +++ /dev/null @@ -1,93 +0,0 @@ -.. _glossary: - -######################## -Paddle文档中使用的词汇表 -######################## - -.. _glossary_paddle: - -PaddlePaddle ------------- - -TBD - -.. _glossary_encode: - -encode ------- - -参考\ :ref:`glossary_encoder`\ 。 - -.. _glossary_encoder: - -encoder -------- - -TBD - -.. _glossary_sample: - -样本 ----- - -TBD Sample的概念 - -.. _glossary_lstm: - -LSTM ----- - -TBD - -.. _glossary_memory: - -Memory ------- - -Memory是 :ref:`glossary_paddle` 实现 :ref:`glossary_RNN` 时候使用的一个概念。 :ref:`glossary_RNN` 即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。 - -.. graphviz:: glossary_rnn.dot - -上图中虚线的连接，即是跨越时间步的网络连接。:ref:`glossary_paddle` 在实现 :ref:`glossary_RNN` 的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的 :ref:`glossary_RNN` 实现便如下图所示。 - -.. graphviz:: glossary_rnn_with_memory.dot - -使用这种方式，:ref:`glossary_paddle` 可以比较简单的判断哪些输出是应该跨越时间步的，哪些不是。 - -.. _glossary_timestep: - -时间步 ------- - -参考 :ref:`_glossary_Sequence` 。 - -.. _glossary_Sequence: - -时间序列 --------- - -时间序列(time series)是指一系列的特征数据。这些特征数据之间的顺序是有意义的。即特征的数组，而不是特征的集合。而这每一个数组元素，或者每一个系列里的特征数据，即为一个时间步(time step)。值得注意的是，时间序列、时间步的概念，并不真正的和『时间』有关。只要一系列特征数据中的『顺序』是有意义的，即为时间序列的输入。 - -举例说明，例如文本分类中，我们通常将一句话理解成一个时间序列。比如一句话中的每一个单词，会变成词表中的位置。而这一句话就可以表示成这些位置的数组。例如 :code:`[9, 2, 3, 5, 3]` 。 - -关于时间序列(time series)的更详细准确的定义，可以参考 `维基百科页面 Time series `_ 或者 `维基百科中文页面 时间序列 `_ 。 - -另外，Paddle中经常会将时间序列成为 :code:`Sequence` 。他们在Paddle的文档和API中是一个概念。 - -.. _glossary_RNN: - -RNN ---- - -RNN 在 :ref:`glossary_paddle` 的文档中，一般表示 :code:`Recurrent neural network`，即时间递归神经网络。详细介绍可以参考 `维基百科页面 Recurrent neural network `_ 或者 `中文维基百科页面 `_ 中关于时间递归神经网络的介绍。 - -RNN 一般在 :ref:`glossary_paddle` 中，指对于一个 :ref:`glossary_Sequence` 输入数据，每一个时间步之间的神经网络具有一定的相关性。例如，某一个神经元的一个输入为上一个时间步网络中某一个神经元的输出。或者，从每一个时间步来看，神经网络的网络结构中具有有向环结构。 - -.. _glossary_双层RNN: - -双层RNN -------- - -双层RNN顾名思义，即 :ref:`glossary_RNN` 之间有一次嵌套关系。输入数据整体上是一个时间序列，而对于每一个内层特征数据而言，也是一个时间序列。即二维数组，或者数组的数组这个概念。 而双层RNN是可以处理这种输入数据的网络结构。 - -例如，对于段落的文本分类，即将一段话进行分类。我们将一段话看成句子的数组，每个句子又是单词的数组。这便是一种双层RNN的输入数据。而将这个段落的每一句话用lstm编码成一个向量，再对每一句话的编码向量用lstm编码成一个段落的向量。再对这个段落向量进行分类，即为这个双层RNN的网络结构。 diff --git a/doc_cn/index.rst b/doc_cn/index.rst index fef39aa527..68128a74f8 100644 --- a/doc_cn/index.rst +++ b/doc_cn/index.rst @@ -11,7 +11,6 @@ PaddlePaddle文档 * `使用示例 `_ * `模型配置 <../doc/ui/api/trainer_config_helpers/index.html>`_ * `集群训练 `_ -* :ref:`glossary` 开发指南 -------- From 0c981164902ea322f5573d645d3bd090b0ce5421 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 23 Nov 2016 17:38:20 +0800 Subject: [PATCH 10/60] Refining documentation --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 43 +++++-------------- 1 file changed, 11 insertions(+), 32 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 896dd7ada9..e1a847fc9c 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -115,13 +115,11 @@ :language: python :lines: 39-66 -- 双层序列，外层memory是单层序列： +.. warning:: + PaddlePaddle目前只支持在每一个时间步中，Memory的sequence长度一致的情况。 - - 由于外层每个时间步返回的是一个子句，这些子句的长度往往不等长。因此当外层有is_seq=True的memory时，内层是**无法直接使用**它的，即内层memory的boot_layer不能链接外层的这个memory。 - - 如果内层memory想**间接使用**这个外层memory，只能通过`pooling_layer`、`last_seq`或`first_seq`这三个layer将它先变成一个元素。但这种情况下，外层memory必须有boot_layer，否则在第0个时间步时，由于外层memory没有任何seq信息，因此上述三个layer的前向会报出“**Check failed: input.sequenceStartPositions**”的错误。 - -示例3：双进双出，输入不等长 -=========================== +示例3：双层RNN，输入不等长 +========================== .. role:: red @@ -129,41 +127,22 @@ -**输入不等长** 是指recurrent_group的多个输入在各时刻的长度可以不相等, 但需要指定一个和输出长度一致的input，用 :red:`targetInlink` 表示。参考配置：单层RNN（:code:`sequence_rnn_multi_unequalength_inputs.conf`），双层RNN（:code:`sequence_nest_rnn_multi_unequalength_inputs.conf`） - -读取双层序列的方法 ------------------- - -我们看一下单双层序列的数据组织形式和dataprovider（见 :code:`rnn_data_provider.py` ） - -.. literalinclude:: ../../../paddle/gserver/tests/rnn_data_provider.py - :language: python - :lines: 69-97 - -data2 中有两个样本，每个样本有两个特征, 记fea1, fea2。 +**输入不等长** 是指recurrent_group的多个输入序列，在每个\ :ref:`glossary_timestep`\ 的子序列长度可以不相等。但\ :ref:`glossary_双层RNN`\ 目前需要整体的输出，与某一个输入的序列信息是一致的。使用\ :red:`targetInlink`\ 可以指定和输出序列信息一致。 -- 单层序列：两个样本分别为[[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]] 和 [[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]] -- 双层序列：两个样本分别为 +本例参考配置分别为\ `单层不等长RNN `_\ 和\ `双层不等长RNN `_\ 。 - - **样本1**\：[[[1, 2], [4, 5, 2]], [[5, 4, 1], [3, 1]]]。fea1和fea2都分别有2个子句，fea1=[[1, 2], [4, 5, 2]], fea2=[[5, 4, 1], [3, 1]] - - **样本2**\：[[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]]。fea1和fea2都分别有3个子句， fea1=[[0, 2], [2, 5], [0, 1, 2]], fea2=[[1, 5], [4], [2, 3, 6, 1]]。
- - **注意**\：每个样本中，各特征的子句数目需要相等。这里说的“双进双出，输入不等长”是指fea1在i时刻的输入的长度可以不等于fea2在i时刻的输入的长度。如对于第1个样本，时刻i=2, fea1[2]=[4, 5, 2]，fea2[2]=[3, 1]，3≠2。 +本例中对于单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 数据完全相同，对于单层\ :ref:`glossary_RNN`\ 的数据一共有两个样本，他们分别是\ :code:`[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]`\ 和\ :code:`[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]`\ 。对于每一个单层\ :ref:`glossary_RNN`\ 的数据，均有两组特征。在单层数据的基础上，\ :ref:`glossary_双层RNN`\ 数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。但是需要注意的是Paddle目前只支持序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。即两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 -- 单双层序列中，两个样本的label都分别是0和1 -模型中的配置 ------------- - -单层RNN（ :code:`sequence_rnn_multi_unequalength_inputs.conf`）和双层RNN（ :code:`v.conf`）两个模型配置达到的效果完全一样，区别只在于输入为单层还是双层序列，现在我们来看它们内部分别是如何实现的。 - -- 单层序列\： +:ref:`glossary_trainer_config`\ 的模型配置 +------------------------------------------ - - 过了一个简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全连接，功能与示例2中`sequence_rnn.conf`的`step`函数完全相同。这里，两个输入x1,x2分别通过calrnn返回最后时刻的状态。结果得到的encoder1_rep和encoder2_rep分别是单层序列，最后取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 - - 注意到这里recurrent_group输入的每个样本中，fea1和fea2的长度都分别相等，这并非偶然，而是因为recurrent_group要求输入为单层序列时，所有输入的长度都必须相等。 +本例中的配置，使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 使用一个\ :code:`recurrent_group`\ 将两个序列同时过完全连接的\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ 的code如下。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf :language: python :lines: 41-58 + :linenos: - 双层序列\： From da7c0f1326677ec7c22b6cdbd6595e2f4a1d59a2 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 23 Nov 2016 17:41:20 +0800 Subject: [PATCH 11/60] Format sequence_nest_rnn_multi_unequalength*.conf --- ...ce_nest_rnn_multi_unequalength_inputs.conf | 106 ----------------- ...ence_nest_rnn_multi_unequalength_inputs.py | 107 ++++++++++++++++++ ...sequence_rnn_multi_unequalength_inputs.py} | 68 +++++------ .../tests/test_RecurrentGradientMachine.cpp | 19 ++-- 4 files changed, 151 insertions(+), 149 deletions(-) delete mode 100644 paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf create mode 100644 paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py rename paddle/gserver/tests/{sequence_rnn_multi_unequalength_inputs.conf => sequence_rnn_multi_unequalength_inputs.py} (52%) diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf deleted file mode 100644 index d0b9450f4b..0000000000 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf +++ /dev/null @@ -1,106 +0,0 @@ -#edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from paddle.trainer_config_helpers import * - -######################## data source ################################ -define_py_data_sources2(train_list='gserver/tests/Sequence/dummy.list', - test_list=None, - module='rnn_data_provider', - obj='process_unequalength_subseq') - - -settings(batch_size=2, learning_rate=0.01) -######################## network configure ################################ -dict_dim = 10 -word_dim = 8 -hidden_dim = 8 -label_dim = 2 - -speaker1 = data_layer(name="word1", size=dict_dim) -speaker2 = data_layer(name="word2", size=dict_dim) - -emb1 = embedding_layer(input=speaker1, size=word_dim) -emb2 = embedding_layer(input=speaker2, size=word_dim) - -# This hierachical RNN is designed to be equivalent to the simple RNN in -# sequence_rnn_multi_unequalength_inputs.conf - -def outer_step(x1, x2): - outer_mem1 = memory(name = "outer_rnn_state1", size = hidden_dim) - outer_mem2 = memory(name = "outer_rnn_state2", size = hidden_dim) - def inner_step1(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem1) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - def inner_step2(y): - inner_mem = memory(name = 'inner_rnn_state_' + y.name, - size = hidden_dim, - boot_layer = outer_mem2) - out = fc_layer(input = [y, inner_mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'inner_rnn_state_' + y.name) - return out - - encoder1 = recurrent_group( - step = inner_step1, - name = 'inner1', - input = x1) - - encoder2 = recurrent_group( - step = inner_step2, - name = 'inner2', - input = x2) - - sentence_last_state1 = last_seq(input = encoder1, name = 'outer_rnn_state1') - sentence_last_state2_ = last_seq(input = encoder2, name = 'outer_rnn_state2') - - encoder1_expand = expand_layer(input = sentence_last_state1, - expand_as = encoder2) - - return [encoder1_expand, encoder2] - - -encoder1_rep, encoder2_rep = recurrent_group( - name="outer", - step=outer_step, - input=[SubsequenceInput(emb1), SubsequenceInput(emb2)], - targetInlink=emb2) - -encoder1_last = last_seq(input = encoder1_rep) -encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) -context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) - -rep = last_seq(input=context) -prob = fc_layer(size=label_dim, - input=rep, - act=SoftmaxActivation(), - bias_attr=True) - -outputs(classification_cost(input=prob, - label=data_layer(name="label", size=label_dim))) - diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py new file mode 100644 index 0000000000..1b709a39c4 --- /dev/null +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py @@ -0,0 +1,107 @@ +#edit-mode: -*- python -*- +# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from paddle.trainer_config_helpers import * + +######################## data source ################################ +define_py_data_sources2( + train_list='gserver/tests/Sequence/dummy.list', + test_list=None, + module='rnn_data_provider', + obj='process_unequalength_subseq') + +settings(batch_size=2, learning_rate=0.01) +######################## network configure ################################ +dict_dim = 10 +word_dim = 8 +hidden_dim = 8 +label_dim = 2 + +speaker1 = data_layer(name="word1", size=dict_dim) +speaker2 = data_layer(name="word2", size=dict_dim) + +emb1 = embedding_layer(input=speaker1, size=word_dim) +emb2 = embedding_layer(input=speaker2, size=word_dim) + +# This hierachical RNN is designed to be equivalent to the simple RNN in +# sequence_rnn_multi_unequalength_inputs.conf + + +def outer_step(x1, x2): + outer_mem1 = memory(name="outer_rnn_state1", size=hidden_dim) + outer_mem2 = memory(name="outer_rnn_state2", size=hidden_dim) + + def inner_step1(y): + inner_mem = memory( + name='inner_rnn_state_' + y.name, + size=hidden_dim, + boot_layer=outer_mem1) + out = fc_layer( + input=[y, inner_mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name='inner_rnn_state_' + y.name) + return out + + def inner_step2(y): + inner_mem = memory( + name='inner_rnn_state_' + y.name, + size=hidden_dim, + boot_layer=outer_mem2) + out = fc_layer( + input=[y, inner_mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name='inner_rnn_state_' + y.name) + return out + + encoder1 = recurrent_group(step=inner_step1, name='inner1', input=x1) + + encoder2 = recurrent_group(step=inner_step2, name='inner2', input=x2) + + sentence_last_state1 = last_seq(input=encoder1, name='outer_rnn_state1') + sentence_last_state2_ = last_seq(input=encoder2, name='outer_rnn_state2') + + encoder1_expand = expand_layer( + input=sentence_last_state1, expand_as=encoder2) + + return [encoder1_expand, encoder2] + + +encoder1_rep, encoder2_rep = recurrent_group( + name="outer", + step=outer_step, + input=[SubsequenceInput(emb1), SubsequenceInput(emb2)], + targetInlink=emb2) + +encoder1_last = last_seq(input=encoder1_rep) +encoder1_expandlast = expand_layer(input=encoder1_last, expand_as=encoder2_rep) +context = mixed_layer( + input=[ + identity_projection(encoder1_expandlast), + identity_projection(encoder2_rep) + ], + size=hidden_dim) + +rep = last_seq(input=context) +prob = fc_layer( + size=label_dim, input=rep, act=SoftmaxActivation(), bias_attr=True) + +outputs( + classification_cost( + input=prob, label=data_layer( + name="label", size=label_dim))) diff --git a/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf b/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py similarity index 52% rename from paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf rename to paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py index 28b1cb98cf..4cf7035477 100644 --- a/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf +++ b/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py @@ -16,11 +16,11 @@ from paddle.trainer_config_helpers import * ######################## data source ################################ -define_py_data_sources2(train_list='gserver/tests/Sequence/dummy.list', - test_list=None, - module='rnn_data_provider', - obj='process_unequalength_seq') - +define_py_data_sources2( + train_list='gserver/tests/Sequence/dummy.list', + test_list=None, + module='rnn_data_provider', + obj='process_unequalength_seq') settings(batch_size=2, learning_rate=0.01) ######################## network configure ################################ @@ -38,38 +38,40 @@ emb2 = embedding_layer(input=speaker2, size=word_dim) # This hierachical RNN is designed to be equivalent to the RNN in # sequence_nest_rnn_multi_unequalength_inputs.conf + def step(x1, x2): - def calrnn(y): - mem = memory(name = 'rnn_state_' + y.name, size = hidden_dim) - out = fc_layer(input = [y, mem], - size = hidden_dim, - act = TanhActivation(), - bias_attr = True, - name = 'rnn_state_' + y.name) - return out - - encoder1 = calrnn(x1) - encoder2 = calrnn(x2) - return [encoder1, encoder2] + def calrnn(y): + mem = memory(name='rnn_state_' + y.name, size=hidden_dim) + out = fc_layer( + input=[y, mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name='rnn_state_' + y.name) + return out + + encoder1 = calrnn(x1) + encoder2 = calrnn(x2) + return [encoder1, encoder2] + encoder1_rep, encoder2_rep = recurrent_group( - name="stepout", - step=step, - input=[emb1, emb2]) + name="stepout", step=step, input=[emb1, emb2]) -encoder1_last = last_seq(input = encoder1_rep) -encoder1_expandlast = expand_layer(input = encoder1_last, - expand_as = encoder2_rep) -context = mixed_layer(input = [identity_projection(encoder1_expandlast), - identity_projection(encoder2_rep)], - size = hidden_dim) +encoder1_last = last_seq(input=encoder1_rep) +encoder1_expandlast = expand_layer(input=encoder1_last, expand_as=encoder2_rep) +context = mixed_layer( + input=[ + identity_projection(encoder1_expandlast), + identity_projection(encoder2_rep) + ], + size=hidden_dim) rep = last_seq(input=context) -prob = fc_layer(size=label_dim, - input=rep, - act=SoftmaxActivation(), - bias_attr=True) - -outputs(classification_cost(input=prob, - label=data_layer(name="label", size=label_dim))) +prob = fc_layer( + size=label_dim, input=rep, act=SoftmaxActivation(), bias_attr=True) +outputs( + classification_cost( + input=prob, label=data_layer( + name="label", size=label_dim))) diff --git a/paddle/gserver/tests/test_RecurrentGradientMachine.cpp b/paddle/gserver/tests/test_RecurrentGradientMachine.cpp index 80d713dac0..9d86067fb5 100644 --- a/paddle/gserver/tests/test_RecurrentGradientMachine.cpp +++ b/paddle/gserver/tests/test_RecurrentGradientMachine.cpp @@ -13,12 +13,12 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include -#include -#include +#include #include #include -#include +#include +#include +#include P_DECLARE_int32(seed); @@ -45,10 +45,9 @@ public: auto p = const_cast(this); auto& params = p->getGradientMachine()->getParameters(); return std::accumulate( - params.begin(), - params.end(), - 0UL, - [](size_t a, const ParameterPtr& p) { return a + p->getSize(); }); + params.begin(), params.end(), 0UL, [](size_t a, const ParameterPtr& p) { + return a + p->getSize(); + }); } }; @@ -148,8 +147,8 @@ TEST(RecurrentGradientMachine, rnn_multi_input) { TEST(RecurrentGradientMachine, rnn_multi_unequalength_input) { for (bool useGpu : {false, true}) { - test("gserver/tests/sequence_rnn_multi_unequalength_inputs.conf", - "gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf", + test("gserver/tests/sequence_rnn_multi_unequalength_inputs.py", + "gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py", 1e-6, useGpu); } From 6aece5060b600f9c86580c5958b6039973a424fa Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 23 Nov 2016 18:06:54 +0800 Subject: [PATCH 12/60] Stash --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 11 +-- ...ence_nest_rnn_multi_unequalength_inputs.py | 69 ++++++++----------- 2 files changed, 36 insertions(+), 44 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index e1a847fc9c..4d29507ca3 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -139,20 +139,21 @@ 本例中的配置，使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 使用一个\ :code:`recurrent_group`\ 将两个序列同时过完全连接的\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ 的code如下。 -.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.conf +.. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py :language: python - :lines: 41-58 + :lines: 42-59 :linenos: - 双层序列\： - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 - - 函数`outer_step`中可以分别处理这两个特征，但我们需要用targetInlink指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 + - 函数`outer_step`中可以分别处理这两个特征，但我们需要用\ :red:`targetInlink`\ 指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 -.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.conf +.. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py :language: python - :lines: 41-89 + :lines: 42-75, 82-89 + :linenos: 示例4：beam_search的生成 ======================== diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py index 1b709a39c4..bf88d00f2d 100644 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py @@ -1,4 +1,4 @@ -#edit-mode: -*- python -*- +# edit-mode: -*- python -*- # Copyright (c) 2016 Baidu, Inc. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); @@ -35,46 +35,37 @@ speaker2 = data_layer(name="word2", size=dict_dim) emb1 = embedding_layer(input=speaker1, size=word_dim) emb2 = embedding_layer(input=speaker2, size=word_dim) -# This hierachical RNN is designed to be equivalent to the simple RNN in -# sequence_rnn_multi_unequalength_inputs.conf - +# This hierarchical RNN is designed to be equivalent to the simple RNN in +# sequence_rnn_multi_unequalength_inputs.conf def outer_step(x1, x2): - outer_mem1 = memory(name="outer_rnn_state1", size=hidden_dim) - outer_mem2 = memory(name="outer_rnn_state2", size=hidden_dim) - - def inner_step1(y): - inner_mem = memory( - name='inner_rnn_state_' + y.name, - size=hidden_dim, - boot_layer=outer_mem1) - out = fc_layer( - input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name='inner_rnn_state_' + y.name) - return out - - def inner_step2(y): - inner_mem = memory( - name='inner_rnn_state_' + y.name, - size=hidden_dim, - boot_layer=outer_mem2) - out = fc_layer( - input=[y, inner_mem], - size=hidden_dim, - act=TanhActivation(), - bias_attr=True, - name='inner_rnn_state_' + y.name) - return out - - encoder1 = recurrent_group(step=inner_step1, name='inner1', input=x1) - - encoder2 = recurrent_group(step=inner_step2, name='inner2', input=x2) - - sentence_last_state1 = last_seq(input=encoder1, name='outer_rnn_state1') - sentence_last_state2_ = last_seq(input=encoder2, name='outer_rnn_state2') + index = [0] + + def inner_step(ipt): + index[0] += 1 + i = index[0] + outer_mem = memory(name="outer_rnn_state_%d" % i, size=hidden_dim) + + def inner_step_impl(y): + inner_mem = memory( + name="inner_rnn_state_" + y.name, + size=hidden_dim, + boot_layer=outer_mem) + out = fc_layer( + input=[y, inner_mem], + size=hidden_dim, + act=TanhActivation(), + bias_attr=True, + name='inner_rnn_state_' + y.name) + return out + + encoder = recurrent_group( + step=inner_step_impl, name='inner_%d' % i, input=ipt) + last = last_seq(name="outer_rnn_state_%d" % i, input=encoder) + return encoder, last + + _, sentence_last_state1 = inner_step(ipt=x1) + encoder2, _ = inner_step(ipt=x2) encoder1_expand = expand_layer( input=sentence_last_state1, expand_as=encoder2) From 24cfc5ab3cb5ee50ff0efe7c5baf5f7d6436ba7a Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Wed, 23 Nov 2016 21:35:27 +0800 Subject: [PATCH 13/60] Minor changes for use_concepts.rst. --- doc_cn/concepts/use_concepts.rst | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/doc_cn/concepts/use_concepts.rst b/doc_cn/concepts/use_concepts.rst index c86429f323..13f6410b98 100644 --- a/doc_cn/concepts/use_concepts.rst +++ b/doc_cn/concepts/use_concepts.rst @@ -15,7 +15,7 @@ PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 系统框图 ======== -下图描述了用户使用框图，PaddlePaddle里链接了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为DataProvider，通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中。下面将分别介绍这两部分。 +下图描述了用户使用框图，PaddlePaddle的trainer进程里内嵌了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为DataProvider，通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中。下面将分别介绍这两部分。 .. graphviz:: @@ -37,17 +37,15 @@ PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 DataProvider ============ -在不同的应用里，训练数据的格式往往各不相同。因此，为了用户能够灵活的处理数据，我们提供了Python处理数据的接口，称为 `PyDataProvider`_ 。 +DataProvider是PaddlePaddle系统的数据提供器，trainer进程会调用DataProvider函数，将用户的原始数据转换成系统可以识别的数据类型。当所有数据读取完一轮后，DataProvider返回空数据，通知系统一轮数据读取结束，系统每一轮训练开始时会重置DataProvider。需要注意的是，DataProvider是被系统调用，而不是新数据驱动系统，一些随机化噪声添加都应该在DataProvider中完成。 -trainer进程会调用DataProvider函数，将用户的原始数据转换成系统可以识别的数据类型。当所有数据读取完一轮后，DataProvider返回空数据，通知系统一轮数据读取结束，系统每一轮训练开始时会重置DataProvider。需要注意的是，DataProvider是被系统调用，而不是新数据驱动系统，一些随机化噪声添加都应该在DataProvider中完成。 - -在 ``PyDataProvider`` 中，系统C++模块接管了shuffle、处理batch、GPU和CPU通信、双缓冲、异步读取等问题，一些情况下(如：``min_pool_size=0``)需要Python接口里处理shuffle，可以参考 `PyDataProvider`_ 的相关文档继续深入了解。 +在不同的应用里，训练数据的格式往往各不相同。因此，为了用户能够灵活的处理数据，我们提供了Python处理数据的接口，称为 `PyDataProvider`_ 。在 ``PyDataProvider`` 中，系统C++模块接管了shuffle、处理batch、GPU和CPU通信、双缓冲、异步读取等问题，一些情况下(如：``min_pool_size=0``)需要Python接口里处理shuffle，可以参考 `PyDataProvider`_ 的相关文档继续深入了解。 模型配置文件 ============ -模型配置主要包括数据传入接口定义(DataConfig)、优化算法(OptimizationConfig)、网络结构(ModelConfig)。 其中数据传入接口定义与DataProvider的关系是：DataProvider里定义数据读取函数，配置文件的DataConfig里指定DataProvider文件名字、生成数据函数接口，请不要混淆。 +模型配置文件主要包括数据传入接口定义(DataConfig)、优化算法(OptimizationConfig)、网络结构(ModelConfig)。 其中数据传入接口定义与DataProvider的关系是：DataProvider里定义数据读取函数，配置文件的DataConfig里指定DataProvider文件名字、生成数据函数接口，请不要混淆。 一个简单的模型配置文件为： @@ -61,7 +59,7 @@ trainer进程会调用DataProvider函数，将用户的原始数据转换成系 DataConfig ---------- -使用函数 ``define_py_data_sources2`` 配置数据源，后缀 2 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 +使用 `PyDataProvider`_ 的函数 ``define_py_data_sources2`` 配置数据源，后缀 2 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 ``define_py_data_sources2`` 里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 @@ -70,7 +68,7 @@ DataConfig OptimizationConfig ------------------ -通过`settings`_ 接口设置神经网络所使用的训练参数和优化算法，包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 +通过`settings`_ 接口设置神经网络所使用的训练参数和 `优化算法`_ ，包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 ModelConfig ----------- @@ -150,6 +148,7 @@ PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trai .. _PyDataProvider: ../ui/data_provider/pydataprovider2.html .. _settings: ../../doc/ui/api/trainer_config_helpers/optimizers.html#settings +.. _优化算法: ../../doc/ui/api/trainer_config_helpers/optimizers.html#optimizers .. _trainer_config_helper: ../../doc/ui/api/trainer_config_helpers/index.html .. _data_layer: ../../doc/ui/api/trainer_config_helpers/layers.html#data-layer .. _simple_img_conv_pool: ../../doc/ui/api/trainer_config_helpers/networks.html#simple-img-conv-pool From 634576128ce3c8075c38dbd7fdf6451ca7885a7f Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Thu, 24 Nov 2016 12:42:22 +0800 Subject: [PATCH 14/60] Done for reviewing docs. --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 4d29507ca3..a13d4728a9 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -144,17 +144,15 @@ :lines: 42-59 :linenos: -- 双层序列\： - - - 双层RNN中，对输入的两个特征分别求时序上的连续全连接(`inner_step1`和`inner_step2`分别处理fea1和fea2)，其功能与示例2中`sequence_nest_rnn.conf`的`outer_step`函数完全相同。不同之处是，此时输入`[SubsequenceInput(emb1), SubsequenceInput(emb2)]`在各时刻并不等长。 - - 函数`outer_step`中可以分别处理这两个特征，但我们需要用\ :red:`targetInlink`\ 指定recurrent_group的输出的格式（各子句长度）只能和其中一个保持一致，如这里选择了和emb2的长度一致。 - - 最后，依然是取encoder1_rep的最后一个时刻和encoder2_rep的所有时刻分别相加得到context。 +而双层序列的代码如下。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py :language: python - :lines: 42-75, 82-89 + :lines: 41-80 :linenos: +在上面代码中，单层和双层序列的使用和示例2中的示例类似，区别是同时处理了两个输入。而对于双层序列，两个输入的子序列长度也并不相同。但是，我们使用了\ :code:`targetInlink`\ 参数设置了外层\ :code:`recurrent_group`\ 的输出格式。所以外层输出的序列形状，和\ :code:`emb2`的序列形状一致。 + 示例4：beam_search的生成 ======================== From 18645134bd07d7a1658c08102d51af8a5d0195cd Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Tue, 29 Nov 2016 16:16:28 +0800 Subject: [PATCH 15/60] Follow comments --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 6 +++--- .../tests/sequence_nest_rnn_multi_unequalength_inputs.py | 4 ++-- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index a13d4728a9..7ae9f5ef8e 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -4,14 +4,14 @@ 单双层RNN API对比介绍 ##################### -这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文中的以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单、双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。文章中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 +这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。本文中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 示例1：双层RNN，子序列间无Memory ================================ 在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_sequence`\ 数据，分别进行序列操作。并且内层的序列操作之间是独立没有依赖的，即不需要使用\ :ref:`glossary_Memory`\ 的。 -在本问题中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用encoder压缩，二者语意上完全一致。这组语意相同的示例配置如下 +在本示例中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用encoder压缩，二者语意上完全一致。这组语意相同的示例配置如下 * 单层\ :ref:`glossary_RNN`\: `sequence_layer_group.conf `_ * :ref:`glossary_双层RNN`\: `sequence_nest_layer_group.conf `_ @@ -22,7 +22,7 @@ 首先，本示例中使用的原始数据如下\: -- 本里中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 +- 本例中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg :language: text diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py index bf88d00f2d..163fce956e 100644 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py @@ -64,8 +64,8 @@ def outer_step(x1, x2): last = last_seq(name="outer_rnn_state_%d" % i, input=encoder) return encoder, last - _, sentence_last_state1 = inner_step(ipt=x1) - encoder2, _ = inner_step(ipt=x2) + encoder1, sentence_last_state1 = inner_step(ipt=x1) + encoder2, sentence_last_state2 = inner_step(ipt=x2) encoder1_expand = expand_layer( input=sentence_last_state1, expand_as=encoder2) From 9d72cab0a4323a6d96bdc443f9cbac5c5658edbc Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Wed, 30 Nov 2016 11:55:00 +0800 Subject: [PATCH 16/60] Accelerating image processing for CNN --- CMakeLists.txt | 4 + plugin/opencv/CMakeLists.txt | 39 ++++++ plugin/opencv/DataTransformer.cpp | 179 +++++++++++++++++++++++++ plugin/opencv/DataTransformer.h | 123 +++++++++++++++++ plugin/opencv/PyDecodejpeg.cpp | 173 ++++++++++++++++++++++++ python/paddle/utils/image_multiproc.py | 170 +++++++++++++++++++++++ python/paddle/utils/image_util.py | 31 +++-- 7 files changed, 705 insertions(+), 14 deletions(-) create mode 100644 plugin/opencv/CMakeLists.txt create mode 100644 plugin/opencv/DataTransformer.cpp create mode 100644 plugin/opencv/DataTransformer.h create mode 100644 plugin/opencv/PyDecodejpeg.cpp create mode 100644 python/paddle/utils/image_multiproc.py diff --git a/CMakeLists.txt b/CMakeLists.txt index af193c27ae..40f18f1550 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -195,3 +195,7 @@ if(WITH_DOC) add_subdirectory(doc) add_subdirectory(doc_cn) endif() + +if(USE_OPENCV) + add_subdirectory(plugin/opencv) +endif() diff --git a/plugin/opencv/CMakeLists.txt b/plugin/opencv/CMakeLists.txt new file mode 100644 index 0000000000..4a253f346a --- /dev/null +++ b/plugin/opencv/CMakeLists.txt @@ -0,0 +1,39 @@ +# use opencv plugin + +project(DeJpeg CXX C) +set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake") +set(PROJ_ROOT ${CMAKE_SOURCE_DIR}) +list(APPEND CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake/Modules) +set(DEJPEG_LINKER_LIBS "") + +# opencv +find_package(OpenCV REQUIRED COMPONENTS core highgui imgproc) +include_directories(${OpenCV_INCLUDE_DIRS}) +list(APPEND DEJPEG_LINKER_LIBS ${OpenCV_LIBS}) +message(STATUS "OpenCV found (${OpenCV_CONFIG_PATH})") +add_definitions(-DUSE_OPENCV) + +# boost-python +set(Boost_NO_SYSTEM_PATHS ON) +if (Boost_NO_SYSTEM_PATHS) + set(BOOST_ROOT $ENV{BOOST_ROOT}) + set(Boost_DIR ${BOOST_ROOT}) + set(Boost_INCLUDE_DIR "${BOOST_ROOT}/include") + set(Boost_LIBRARIES "${BOOST_ROOT}/lib/") +endif (Boost_NO_SYSTEM_PATHS) +find_package(Boost 1.46 COMPONENTS python) +include_directories(SYSTEM ${Boost_INCLUDE_DIR}) +link_directories(${Boost_INCLUDE_DIR}) +message(STATUS "Boost found (${Boost_INCLUDE_DIR})") +message(STATUS "Boost found (${Boost_LIBRARIES})") +list(APPEND DEJPEG_LINKER_LIBS ${Boost_LIBRARIES}) + + +file(GLOB DEJPEG_HEADER "${CMAKE_CURRENT_SOURCE_DIR}" "*.h") +file(GLOB DEJPEG_SOURCES "${CMAKE_CURRENT_SOURCE_DIR}" "*.cpp") + +set(CMAKE_CXX_FLAGS "-std=c++11 -O3 -fPIC -Wno-unused-parameter") + +add_library(DeJpeg SHARED ${DEJPEG_SOURCES}) +target_link_libraries(DeJpeg ${DEJPEG_LINKER_LIBS}) +set_target_properties(DeJpeg PROPERTIES PREFIX "") diff --git a/plugin/opencv/DataTransformer.cpp b/plugin/opencv/DataTransformer.cpp new file mode 100644 index 0000000000..f4e21db886 --- /dev/null +++ b/plugin/opencv/DataTransformer.cpp @@ -0,0 +1,179 @@ +/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. + +Licensed under the Apache License, Version 2.0 (the "License"); +you may not use this file except in compliance with the License. +You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. */ + +#include "DataTransformer.h" +#include +#include + +DataTransformer::DataTransformer(int threadNum, + int capacity, + bool isTest, + bool isColor, + int cropHeight, + int cropWidth, + int imgSize, + bool isEltMean, + bool isChannelMean, + float* meanValues) + : isTest_(isTest), + isColor_(isColor), + cropHeight_(cropHeight), + cropWidth_(cropWidth), + imgSize_(imgSize), + capacity_(capacity), + prefetchFree_(capacity), + prefetchFull_(capacity) { + fetchCount_ = -1; + scale_ = 1.0; + isChannelMean_ = isChannelMean; + isEltMean_ = isEltMean; + loadMean(meanValues); + + imgPixels_ = cropHeight * cropWidth * (isColor_ ? 3 : 1); + + prefetch_.reserve(capacity); + for (int i = 0; i < capacity; i++) { + auto d = std::make_shared(new float[imgPixels_ * 3], 0); + prefetch_.push_back(d); + memset(prefetch_[i]->first, 0, imgPixels_ * sizeof(float)); + prefetchFree_.enqueue(prefetch_[i]); + } + + numThreads_ = 12; + syncThreadPool_.reset(new SyncThreadPool(numThreads_, false)); +} + +void DataTransformer::loadMean(float* values) { + if (values) { + int c = isColor_ ? 3 : 1; + int sz = isChannelMean_ ? c : cropHeight_ * cropWidth_ * c; + meanValues_ = new float[sz]; + memcpy(meanValues_, values, sz * sizeof(float)); + } +} + +void DataTransformer::startFetching(const char* src, + const int size, + float* trg) { + vector imbuf(src, src + size); + int cvFlag = (isColor_ ? CV_LOAD_IMAGE_COLOR : CV_LOAD_IMAGE_GRAYSCALE); + cv::Mat im = cv::imdecode(cv::Mat(imbuf), cvFlag); + if (!im.data) { + LOG(ERROR) << "Could not decode image"; + LOG(ERROR) << im.channels() << " " << im.rows << " " << im.cols; + } + this->transform(im, trg); +} + +int DataTransformer::Rand(int min, int max) { + std::random_device source; + std::mt19937 rng(source()); + std::uniform_int_distribution dist(min, max); + return dist(rng); +} + +void DataTransformer::transform(Mat& cvImgOri, float* target) { + const int imgChannels = cvImgOri.channels(); + const int imgHeight = cvImgOri.rows; + const int imgWidth = cvImgOri.cols; + const bool doMirror = (!isTest_) && Rand(0, 1); + int h_off = 0; + int w_off = 0; + int th = imgHeight; + int tw = imgWidth; + cv::Mat img; + if (imgSize_ > 0) { + if (imgHeight > imgWidth) { + tw = imgSize_; + th = int(double(imgHeight) / imgWidth * tw); + th = th > imgSize_ ? th : imgSize_; + } else { + th = imgSize_; + tw = int(double(imgWidth) / imgHeight * th); + tw = tw > imgSize_ ? tw : imgSize_; + } + cv::resize(cvImgOri, img, cv::Size(tw, th)); + } else { + cv::Mat img = cvImgOri; + } + + cv::Mat cv_cropped_img = img; + if (cropHeight_ && cropWidth_) { + if (!isTest_) { + h_off = Rand(0, th - cropHeight_); + w_off = Rand(0, tw - cropWidth_); + } else { + h_off = (th - cropHeight_) / 2; + w_off = (tw - cropWidth_) / 2; + } + cv::Rect roi(w_off, h_off, cropWidth_, cropHeight_); + cv_cropped_img = img(roi); + } else { + CHECK_EQ(cropHeight_, imgHeight); + CHECK_EQ(cropWidth_, imgWidth); + } + int height = cropHeight_; + int width = cropWidth_; + int top_index; + for (int h = 0; h < height; ++h) { + const uchar* ptr = cv_cropped_img.ptr(h); + int img_index = 0; + for (int w = 0; w < width; ++w) { + for (int c = 0; c < imgChannels; ++c) { + if (doMirror) { + top_index = (c * height + h) * width + width - 1 - w; + } else { + top_index = (c * height + h) * width + w; + } + float pixel = static_cast(ptr[img_index++]); + if (isEltMean_) { + int mean_index = (c * imgHeight + h) * imgWidth + w; + target[top_index] = (pixel - meanValues_[mean_index]) * scale_; + } else { + if (isChannelMean_) { + target[top_index] = (pixel - meanValues_[c]) * scale_; + } else { + target[top_index] = pixel * scale_; + } + } + } + } + } // target: BGR +} + +void DataTransformer::start(vector& data, int* datalen, int* labels) { + auto job = [&](int tid, int numThreads) { + for (int i = tid; i < data.size(); i += numThreads) { + DataTypePtr ret = prefetchFree_.dequeue(); + char* buf = data[i]; + int size = datalen[i]; + ret->second = labels[i]; + this->startFetching(buf, size, ret->first); + prefetchFull_.enqueue(ret); + } + }; + syncThreadPool_->exec(job); + fetchCount_ = data.size(); +} + +void DataTransformer::obtain(float* data, int* label) { + fetchCount_--; + if (fetchCount_ < 0) { + LOG(FATAL) << "Empty data"; + } + DataTypePtr ret = prefetchFull_.dequeue(); + *label = ret->second; + memcpy(data, ret->first, sizeof(float) * imgPixels_); + prefetchFree_.enqueue(ret); +} diff --git a/plugin/opencv/DataTransformer.h b/plugin/opencv/DataTransformer.h new file mode 100644 index 0000000000..c4f04a5878 --- /dev/null +++ b/plugin/opencv/DataTransformer.h @@ -0,0 +1,123 @@ +/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. + +Licensed under the Apache License, Version 2.0 (the "License"); +you may not use this file except in compliance with the License. +You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. */ + +#include +#include +// #define OPENCV_CAN_BREAK_BINARY_COMPATIBILITY +#include +#include +#include +#include + +#include "paddle/utils/Thread.h" + +using namespace std; +using namespace cv; +using namespace paddle; + +/** + * This is an image processing module with OpenCV, such as + * resizing, scaling, mirroring, substracting the image mean... + * + * This class has a double BlockQueue and they shared the same memory. + * It is used to avoid create memory each time. And it also can + * return the data even if the data are processing in multi-threads. + */ +class DataTransformer { +public: + DataTransformer(int threadNum, + int capacity, + bool isTest, + bool isColor, + int cropHeight, + int cropWidth, + int imgSize, + bool isEltMean, + bool isChannelMean, + float* meanValues); + virtual ~DataTransformer() { + if (meanValues_) { + free(meanValues_); + } + } + + /** + * @brief Start multi-threads to transform a list of input data. + * The processed data will be saved in Queue of prefetchFull_. + * + * @param data Data containing the image string to be transformed. + * @param label The label of input image. + */ + void start(vector& data, int* datalen, int* labels); + + /** + * @brief Applies the transformation on one image Mat. + * + * @param img The input img to be transformed. + * @param target target is used to save the transformed data. + */ + void transform(Mat& img, float* target); + + /** + * @brief Decode the image string, then calls transform() function. + * + * @param src The input image string. + * @param size The length of string. + * @param trg trg is used to save the transformed data. + */ + void startFetching(const char* src, const int size, float* trg); + + /** + * @brief Return the transformed data and its label. + */ + void obtain(float* data, int* label); + +private: + int isTest_; + int isColor_; + int cropHeight_; + int cropWidth_; + int imgSize_; + int capacity_; + int fetchCount_; + bool isEltMean_; + bool isChannelMean_; + int numThreads_; + float scale_; + int imgPixels_; + float* meanValues_; + + /** + * Initialize the mean values. + */ + void loadMean(float* values); + + /** + * @brief Generates a random integer from Uniform({min, min + 1, ..., max}). + * @param min The lower bound (inclusive) value of the random number. + * @param max The upper bound (inclusive) value of the random number. + * + * @return + * A uniformly random integer value from ({min, min + 1, ..., max}). + */ + int Rand(int min, int max); + + typedef pair DataType; + typedef std::shared_ptr DataTypePtr; + std::vector prefetch_; + std::unique_ptr syncThreadPool_; + BlockingQueue prefetchFree_; + BlockingQueue prefetchFull_; + +}; // class DataTransformer diff --git a/plugin/opencv/PyDecodejpeg.cpp b/plugin/opencv/PyDecodejpeg.cpp new file mode 100644 index 0000000000..b004d7cad8 --- /dev/null +++ b/plugin/opencv/PyDecodejpeg.cpp @@ -0,0 +1,173 @@ +/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. + +Licensed under the Apache License, Version 2.0 (the "License"); +you may not use this file except in compliance with the License. +You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. */ + +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "DataTransformer.h" + +using namespace boost::python; +using namespace std; + +/** + * DecodeJpeg is an image processing API for interfacing Python and C++ + * code DataTransformer, which used OpenCV and multi-threads to accelerate + * image processing. + * The Boost Python Library is used to wrap C++ interfaces. + */ + +class DecodeJpeg { +public: + /** + * The constructor will create and nitialize an object of DataTransformer. + */ + DecodeJpeg(int threadNum, + int capacity, + bool isTest, + bool isColor, + int resize_min_size, + int cropSizeH, + int cropSizeW, + PyObject* meanValues) { + int channel = isColor ? 3 : 1; + bool isEltMean = false; + bool isChannelMean = false; + float* mean = NULL; + if (meanValues || meanValues != Py_None) { + if (!PyArray_Check(meanValues)) { + LOG(FATAL) << "Object is not a numpy array"; + } + pyTypeCheck(meanValues); + int size = PyArray_SIZE(meanValues); + isChannelMean = (size == channel) ? true : false; + isEltMean = (size == channel * cropSizeH * cropSizeW) ? true : false; + CHECK(isChannelMean != isEltMean); + mean = (float*)PyArray_DATA(meanValues); + } + tfhandlerPtr_ = std::make_shared(threadNum, + capacity, + isTest, + isColor, + cropSizeH, + cropSizeW, + resize_min_size, + isEltMean, + isChannelMean, + mean); + } + + ~DecodeJpeg() {} + + /** + * @brief This function is used to parse the Python object and convert + * the data to C++ format. Then it called the function of + * DataTransformer to start image processing. + * @param pysrc The input image list with string type. + * @param pylabel The input label of image. + * It's type is numpy.array with int32. + */ + void start(boost::python::list& pysrc, PyObject* pydlen, PyObject* pylabel) { + vector data; + int num = len(pysrc); + for (int t = 0; t < num; ++t) { + char* src = boost::python::extract(pysrc[t]); + data.push_back(src); + } + int* dlen = (int*)PyArray_DATA(pydlen); + int* dlabels = (int*)PyArray_DATA(pylabel); + tfhandlerPtr_->start(data, dlen, dlabels); + } + + /** + * @brief Return one processed data. + * @param pytrg The processed image. + * @param pylabel The label of processed image. + */ + void get(PyObject* pytrg, PyObject* pylab) { + pyWritableCheck(pytrg); + pyWritableCheck(pylab); + pyContinuousCheck(pytrg); + pyContinuousCheck(pylab); + float* data = (float*)PyArray_DATA(pytrg); + int* label = (int*)PyArray_DATA(pylab); + tfhandlerPtr_->obtain(data, label); + } + + /** + * @brief An object of DataTransformer, which is used to call + * the image processing funtions. + */ + std::shared_ptr tfhandlerPtr_; + +private: + /** + * @brief Check whether the type of PyObject is valid or not. + */ + void pyTypeCheck(const PyObject* o) { + int typenum = PyArray_TYPE(o); + + // clang-format off + int type = + typenum == NPY_UBYTE ? CV_8U : + typenum == NPY_BYTE ? CV_8S : + typenum == NPY_USHORT ? CV_16U : + typenum == NPY_SHORT ? CV_16S : + typenum == NPY_INT || typenum == NPY_LONG ? CV_32S : + typenum == NPY_FLOAT ? CV_32F : + typenum == NPY_DOUBLE ? CV_64F : -1; + // clang-format on + + if (type < 0) { + LOG(FATAL) << "toMat: Data type = " << type << " is not supported"; + } + } + + /** + * @brief Check whether the PyObject is writable or not. + */ + void pyWritableCheck(PyObject* o) { CHECK(PyArray_ISWRITEABLE(o)); } + + /** + * @brief Check whether the PyObject is c-contiguous or not. + */ + void pyContinuousCheck(PyObject* o) { CHECK(PyArray_IS_C_CONTIGUOUS(o)); } +}; + +/** + * @brief Initialize the Python interpreter and numpy. + */ +static void initPython() { + Py_Initialize(); + PyOS_sighandler_t sighandler = PyOS_getsig(SIGINT); + import_array(); + PyOS_setsig(SIGINT, sighandler); +} + +/** + * Use Boost.Python to expose C++ interface to Python. + */ +BOOST_PYTHON_MODULE(DeJpeg) { + initPython(); + class_("DecodeJpeg", + init()) + .def("start", &DecodeJpeg::start) + .def("get", &DecodeJpeg::get); +}; diff --git a/python/paddle/utils/image_multiproc.py b/python/paddle/utils/image_multiproc.py new file mode 100644 index 0000000000..ccc0a531a7 --- /dev/null +++ b/python/paddle/utils/image_multiproc.py @@ -0,0 +1,170 @@ +import os, psutil +import cv2 +from paddle.utils.image_util import * +import multiprocessing +import subprocess, signal, sys + + +class CvImageTransfomer(ImageTransformer): + """ + CvImageTransfomer used python-opencv to process image. + """ + + def __init__(self, + min_size=None, + crop_size=None, + transpose=None, + channel_swap=None, + mean=None, + is_train=True, + is_color=True): + ImageTransformer.__init__(self, transpose, channel_swap, mean, is_color) + self.min_size = min_size + self.crop_size = crop_size + self.is_train = is_train + + def cv_resize_fixed_short_side(self, im, min_size): + row, col = im.shape[:2] + scale = min_size / float(min(row, col)) + if row < col: + row = min_size + col = int(round(col * scale)) + col = col if col > min_size else min_size + else: + col = min_size + row = int(round(row * scale)) + row = row if row > min_size else min_size + resized_size = row, col + im = cv2.resize(im, resized_size, interpolation=cv2.INTER_CUBIC) + return im + + def crop_img(self, im): + """ + Return cropped image. + The size of the cropped image is inner_size * inner_size. + im: (H x W x K) ndarrays + """ + row, col = im.shape[:2] + start_h, start_w = 0, 0 + if self.is_train: + start_h = np.random.randint(0, row - self.crop_size + 1) + start_w = np.random.randint(0, col - self.crop_size + 1) + else: + start_h = (row - self.crop_size) / 2 + start_w = (col - self.crop_size) / 2 + end_h, end_w = start_h + self.crop_size, start_w + self.crop_size + if self.is_color: + im = im[start_h:end_h, start_w:end_w, :] + else: + im = im[start_h:end_h, start_w:end_w] + if (self.is_train) and (np.random.randint(2) == 0): + if self.is_color: + im = im[:, ::-1, :] + else: + im = im[:, ::-1] + return im + + def transform(self, im): + im = self.cv_resize_fixed_short_side(im, self.min_size) + im = self.crop_img(im) + # transpose, swap channel, sub mean + im = im.astype('float32') + ImageTransformer.transformer(self, im) + return im + + def load_image_from_string(self, data): + flag = cv2.CV_LOAD_IMAGE_COLOR if self.is_color else cv2.CV_LOAD_IMAGE_GRAYSCALE + im = cv2.imdecode(np.fromstring(data, np.uint8), flag) + return im + + def transform_from_string(self, data): + im = self.load_image_from_string(data) + return self.transform(im) + + +class MultiProcessImageTransfomer(): + def __init__(self, + procnum=10, + capacity=10240, + min_size=None, + crop_size=None, + transpose=None, + channel_swap=None, + mean=None, + is_train=True, + is_color=True): + self.procnum = procnum + self.capacity = capacity + self.size = 0 + self.count = 0 + signal.signal(signal.SIGTERM, self.kill_child_processes) + self.fetch_queue = multiprocessing.Queue(maxsize=capacity) + self.cv_transformer = CvImageTransfomer(min_size, crop_size, transpose, + channel_swap, mean, is_train, + is_color) + + def __del__(self): + try: + for p in self.procs: + p.join() + except Exception as e: + print str(e) + + def reset(self, size): + self.size = size + self.count = 0 + self.procs = [] + + def run_proc(self, data, label): + dlen = len(label) + self.reset(dlen) + for i in xrange(self.procnum): + start = dlen * i / self.procnum + end = dlen * (i + 1) / self.procnum + proc = multiprocessing.Process( + target=self.batch_transfomer, + args=(data[start:end], label[start:end])) + proc.daemon = True + self.procs.append(proc) + for p in self.procs: + p.start() + + def get(self): + """ + Return one processed image. + """ + # block if necessary until an item is available + data, lab = self.fetch_queue.get(block=True) + self.count += 1 + if self.count == self.size: + try: + for p in self.procs: + p.join() + except Exception as e: + print str(e) + return data, lab + + def batch_transfomer(self, data, label): + """ + param data: input data in format of image string + type data: a list of string + label: the label of image + """ + for i in xrange(len(label)): + res = self.cv_transformer.transform_from_string(data[i]) + self.fetch_queue.put((res, int(label[i]))) + + def kill_child_processes(self, signum, frame): + """ + Kill a process's child processes in python. + """ + parent_id = os.getpid() + ps_command = subprocess.Popen( + "ps -o pid --ppid %d --noheaders" % parent_id, + shell=True, + stdout=subprocess.PIPE) + ps_output = ps_command.stdout.read() + retcode = ps_command.wait() + for pid_str in ps_output.strip().split("\n")[:-1]: + os.kill(int(pid_str), signal.SIGTERM) + sys.exit() diff --git a/python/paddle/utils/image_util.py b/python/paddle/utils/image_util.py index b5c6431c06..e6c6b04de0 100644 --- a/python/paddle/utils/image_util.py +++ b/python/paddle/utils/image_util.py @@ -186,29 +186,32 @@ class ImageTransformer: channel_swap=None, mean=None, is_color=True): - self.transpose = transpose - self.channel_swap = None - self.mean = None self.is_color = is_color + self.set_transpose(transpose) + self.set_channel_swap(channel_swap) + self.set_mean(mean) def set_transpose(self, order): - if self.is_color: - assert 3 == len(order) + if order is not None: + if self.is_color: + assert 3 == len(order) self.transpose = order def set_channel_swap(self, order): - if self.is_color: - assert 3 == len(order) + if order is not None: + if self.is_color: + assert 3 == len(order) self.channel_swap = order def set_mean(self, mean): - # mean value, may be one value per channel - if mean.ndim == 1: - mean = mean[:, np.newaxis, np.newaxis] - else: - # elementwise mean - if self.is_color: - assert len(mean.shape) == 3 + if mean is not None: + # mean value, may be one value per channel + if mean.ndim == 1: + mean = mean[:, np.newaxis, np.newaxis] + else: + # elementwise mean + if self.is_color: + assert len(mean.shape) == 3 self.mean = mean def transformer(self, data): From fe073d1f2a81c37c22ddc245722014c13e78e7af Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Wed, 30 Nov 2016 20:01:42 +0800 Subject: [PATCH 17/60] Add style check and remove 'using namespace' --- CMakeLists.txt | 1 + plugin/opencv/CMakeLists.txt | 3 +++ plugin/opencv/DataTransformer.cpp | 10 ++++++---- plugin/opencv/DataTransformer.h | 19 ++++++++++--------- plugin/opencv/PyDecodejpeg.cpp | 9 ++++----- 5 files changed, 24 insertions(+), 18 deletions(-) diff --git a/CMakeLists.txt b/CMakeLists.txt index fcb8398517..c9cdfc2c2b 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -43,6 +43,7 @@ option(WITH_SWIG_PY "Compile PaddlePaddle with py PaddlePaddle prediction api" $ option(ON_TRAVIS "Running test on travis-ci or not." OFF) option(ON_COVERALLS "Generating code coverage data on coveralls or not." OFF) option(COVERALLS_UPLOAD "Uploading the generated coveralls json." ON) +option(USE_OPENCV "Compile PaddlePaddle with opencv" OFF) if(NOT CMAKE_BUILD_TYPE) set(CMAKE_BUILD_TYPE "RelWithDebInfo" CACHE STRING diff --git a/plugin/opencv/CMakeLists.txt b/plugin/opencv/CMakeLists.txt index bc0a6e6354..7a6b22c899 100644 --- a/plugin/opencv/CMakeLists.txt +++ b/plugin/opencv/CMakeLists.txt @@ -42,3 +42,6 @@ add_library(DeJpeg SHARED ${DEJPEG_SOURCES}) target_compile_options(DeJpeg BEFORE PRIVATE ${BUILD_PRIVATE_FLAGS}) target_link_libraries(DeJpeg ${DEJPEG_LINKER_LIBS}) set_target_properties(DeJpeg PROPERTIES PREFIX "") + +add_style_check_target(DeJpeg ${DEJPEG_SOURCES}) +add_style_check_target(DeJpeg ${DEJPEG_HEADER}) diff --git a/plugin/opencv/DataTransformer.cpp b/plugin/opencv/DataTransformer.cpp index d9e8883443..dd123639f4 100644 --- a/plugin/opencv/DataTransformer.cpp +++ b/plugin/opencv/DataTransformer.cpp @@ -51,7 +51,7 @@ DataTransformer::DataTransformer(int threadNum, } numThreads_ = threadNum; - syncThreadPool_.reset(new SyncThreadPool(numThreads_, false)); + syncThreadPool_.reset(new paddle::SyncThreadPool(numThreads_, false)); } void DataTransformer::loadMean(float* values) { @@ -66,7 +66,7 @@ void DataTransformer::loadMean(float* values) { void DataTransformer::startFetching(const char* src, const int size, float* trg) { - vector imbuf(src, src + size); + std::vector imbuf(src, src + size); int cvFlag = (isColor_ ? CV_LOAD_IMAGE_COLOR : CV_LOAD_IMAGE_GRAYSCALE); cv::Mat im = cv::imdecode(cv::Mat(imbuf), cvFlag); if (!im.data) { @@ -83,7 +83,7 @@ int DataTransformer::Rand(int min, int max) { return dist(rng); } -void DataTransformer::transform(Mat& cvImgOri, float* target) { +void DataTransformer::transform(cv::Mat& cvImgOri, float* target) { const int imgChannels = cvImgOri.channels(); const int imgHeight = cvImgOri.rows; const int imgWidth = cvImgOri.cols; @@ -152,7 +152,9 @@ void DataTransformer::transform(Mat& cvImgOri, float* target) { } // target: BGR } -void DataTransformer::start(vector& data, int* datalen, int* labels) { +void DataTransformer::start(std::vector& data, + int* datalen, + int* labels) { auto job = [&](int tid, int numThreads) { for (size_t i = tid; i < data.size(); i += numThreads) { DataTypePtr ret = prefetchFree_.dequeue(); diff --git a/plugin/opencv/DataTransformer.h b/plugin/opencv/DataTransformer.h index 52abab928b..603cea3059 100644 --- a/plugin/opencv/DataTransformer.h +++ b/plugin/opencv/DataTransformer.h @@ -12,6 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#ifndef DATATRANSFORMER_H_ +#define DATATRANSFORMER_H_ + #include #include #include @@ -21,9 +24,6 @@ limitations under the License. */ #include "paddle/utils/Thread.h" -using namespace cv; -using namespace paddle; - /** * This is an image processing module with OpenCV, such as * resizing, scaling, mirroring, substracting the image mean... @@ -57,7 +57,7 @@ public: * @param data Data containing the image string to be transformed. * @param label The label of input image. */ - void start(vector& data, int* datalen, int* labels); + void start(std::vector& data, int* datalen, int* labels); /** * @brief Applies the transformation on one image Mat. @@ -65,7 +65,7 @@ public: * @param img The input img to be transformed. * @param target target is used to save the transformed data. */ - void transform(Mat& img, float* target); + void transform(cv::Mat& img, float* target); /** * @brief Decode the image string, then calls transform() function. @@ -114,8 +114,9 @@ private: typedef std::pair DataType; typedef std::shared_ptr DataTypePtr; std::vector prefetch_; - std::unique_ptr syncThreadPool_; - BlockingQueue prefetchFree_; - BlockingQueue prefetchFull_; - + std::unique_ptr syncThreadPool_; + paddle::BlockingQueue prefetchFree_; + paddle::BlockingQueue prefetchFull_; }; // class DataTransformer + +#endif // DATATRANSFORMER_H_ diff --git a/plugin/opencv/PyDecodejpeg.cpp b/plugin/opencv/PyDecodejpeg.cpp index 66054302f8..a32e6430e1 100644 --- a/plugin/opencv/PyDecodejpeg.cpp +++ b/plugin/opencv/PyDecodejpeg.cpp @@ -23,8 +23,6 @@ limitations under the License. */ #include "DataTransformer.h" -using namespace boost::python; - /** * DecodeJpeg is an image processing API for interfacing Python and C++ * code DataTransformer, which used OpenCV and multi-threads to accelerate @@ -83,7 +81,7 @@ public: * It's type is numpy.array with int32. */ void start(boost::python::list& pysrc, PyObject* pydlen, PyObject* pylabel) { - vector data; + std::vector data; int num = len(pysrc); for (int t = 0; t < num; ++t) { char* src = boost::python::extract(pysrc[t]); @@ -169,8 +167,9 @@ static void initPython() { */ BOOST_PYTHON_MODULE(DeJpeg) { initPython(); - class_("DecodeJpeg", - init()) + boost::python::class_( + "DecodeJpeg", + boost::python::init()) .def("start", &DecodeJpeg::start) .def("get", &DecodeJpeg::get); }; From ae06debf2348d5f4df4aeeff8f99e89d31ab30a6 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Fri, 2 Dec 2016 15:09:27 +0800 Subject: [PATCH 18/60] Remove the C++ code and refine Python code. --- CMakeLists.txt | 5 - plugin/opencv/CMakeLists.txt | 47 ---- plugin/opencv/DataTransformer.cpp | 181 -------------- plugin/opencv/DataTransformer.h | 122 ---------- plugin/opencv/PyDecodejpeg.cpp | 175 -------------- python/paddle/utils/image_multiproc.py | 313 ++++++++++++++++--------- 6 files changed, 208 insertions(+), 635 deletions(-) delete mode 100644 plugin/opencv/CMakeLists.txt delete mode 100644 plugin/opencv/DataTransformer.cpp delete mode 100644 plugin/opencv/DataTransformer.h delete mode 100644 plugin/opencv/PyDecodejpeg.cpp diff --git a/CMakeLists.txt b/CMakeLists.txt index c9cdfc2c2b..7d685587a7 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -43,7 +43,6 @@ option(WITH_SWIG_PY "Compile PaddlePaddle with py PaddlePaddle prediction api" $ option(ON_TRAVIS "Running test on travis-ci or not." OFF) option(ON_COVERALLS "Generating code coverage data on coveralls or not." OFF) option(COVERALLS_UPLOAD "Uploading the generated coveralls json." ON) -option(USE_OPENCV "Compile PaddlePaddle with opencv" OFF) if(NOT CMAKE_BUILD_TYPE) set(CMAKE_BUILD_TYPE "RelWithDebInfo" CACHE STRING @@ -196,7 +195,3 @@ if(WITH_DOC) add_subdirectory(doc) add_subdirectory(doc_cn) endif() - -if(USE_OPENCV) - add_subdirectory(plugin/opencv) -endif() diff --git a/plugin/opencv/CMakeLists.txt b/plugin/opencv/CMakeLists.txt deleted file mode 100644 index 7a6b22c899..0000000000 --- a/plugin/opencv/CMakeLists.txt +++ /dev/null @@ -1,47 +0,0 @@ -# use opencv plugin - -project(DeJpeg CXX C) -set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake") -set(PROJ_ROOT ${CMAKE_SOURCE_DIR}) -list(APPEND CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake/Modules) -set(DEJPEG_LINKER_LIBS "") - -# opencv -find_package(OpenCV REQUIRED COMPONENTS core highgui imgproc) -include_directories(${OpenCV_INCLUDE_DIRS}) -list(APPEND DEJPEG_LINKER_LIBS ${OpenCV_LIBS}) -message(STATUS "OpenCV found (${OpenCV_CONFIG_PATH})") -add_definitions(-DUSE_OPENCV) - -# boost-python -set(Boost_NO_SYSTEM_PATHS ON) -if (Boost_NO_SYSTEM_PATHS) - set(BOOST_ROOT $ENV{BOOST_ROOT}) - set(Boost_DIR ${BOOST_ROOT}) - set(Boost_INCLUDE_DIR "${BOOST_ROOT}/include") - set(Boost_LIBRARIES "${BOOST_ROOT}/lib/") -endif (Boost_NO_SYSTEM_PATHS) -find_package(Boost 1.46 COMPONENTS python) -include_directories(SYSTEM ${Boost_INCLUDE_DIR}) -link_directories(${Boost_INCLUDE_DIR}) -message(STATUS "Boost found (${Boost_INCLUDE_DIR})") -message(STATUS "Boost found (${Boost_LIBRARIES})") -list(APPEND DEJPEG_LINKER_LIBS ${Boost_LIBRARIES}) - - -file(GLOB DEJPEG_HEADER "${CMAKE_CURRENT_SOURCE_DIR}" "*.h") -file(GLOB DEJPEG_SOURCES "${CMAKE_CURRENT_SOURCE_DIR}" "*.cpp") - -set(BUILD_PRIVATE_FLAGS - -Wno-all - -Wno-error - -Wno-non-virtual-dtor - -Wno-delete-non-virtual-dtor) - -add_library(DeJpeg SHARED ${DEJPEG_SOURCES}) -target_compile_options(DeJpeg BEFORE PRIVATE ${BUILD_PRIVATE_FLAGS}) -target_link_libraries(DeJpeg ${DEJPEG_LINKER_LIBS}) -set_target_properties(DeJpeg PROPERTIES PREFIX "") - -add_style_check_target(DeJpeg ${DEJPEG_SOURCES}) -add_style_check_target(DeJpeg ${DEJPEG_HEADER}) diff --git a/plugin/opencv/DataTransformer.cpp b/plugin/opencv/DataTransformer.cpp deleted file mode 100644 index dd123639f4..0000000000 --- a/plugin/opencv/DataTransformer.cpp +++ /dev/null @@ -1,181 +0,0 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. - -Licensed under the Apache License, Version 2.0 (the "License"); -you may not use this file except in compliance with the License. -You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - -Unless required by applicable law or agreed to in writing, software -distributed under the License is distributed on an "AS IS" BASIS, -WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -See the License for the specific language governing permissions and -limitations under the License. */ - -#include "DataTransformer.h" -#include -#include - -DataTransformer::DataTransformer(int threadNum, - int capacity, - bool isTest, - bool isColor, - int cropHeight, - int cropWidth, - int imgSize, - bool isEltMean, - bool isChannelMean, - float* meanValues) - : isTest_(isTest), - isColor_(isColor), - cropHeight_(cropHeight), - cropWidth_(cropWidth), - imgSize_(imgSize), - capacity_(capacity), - prefetchFree_(capacity), - prefetchFull_(capacity) { - fetchCount_ = -1; - scale_ = 1.0; - isChannelMean_ = isChannelMean; - isEltMean_ = isEltMean; - loadMean(meanValues); - - imgPixels_ = cropHeight * cropWidth * (isColor_ ? 3 : 1); - - prefetch_.reserve(capacity); - for (int i = 0; i < capacity; i++) { - auto d = std::make_shared(new float[imgPixels_ * 3], 0); - prefetch_.push_back(d); - memset(prefetch_[i]->first, 0, imgPixels_ * sizeof(float)); - prefetchFree_.enqueue(prefetch_[i]); - } - - numThreads_ = threadNum; - syncThreadPool_.reset(new paddle::SyncThreadPool(numThreads_, false)); -} - -void DataTransformer::loadMean(float* values) { - if (values) { - int c = isColor_ ? 3 : 1; - int sz = isChannelMean_ ? c : cropHeight_ * cropWidth_ * c; - meanValues_ = new float[sz]; - memcpy(meanValues_, values, sz * sizeof(float)); - } -} - -void DataTransformer::startFetching(const char* src, - const int size, - float* trg) { - std::vector imbuf(src, src + size); - int cvFlag = (isColor_ ? CV_LOAD_IMAGE_COLOR : CV_LOAD_IMAGE_GRAYSCALE); - cv::Mat im = cv::imdecode(cv::Mat(imbuf), cvFlag); - if (!im.data) { - LOG(ERROR) << "Could not decode image"; - LOG(ERROR) << im.channels() << " " << im.rows << " " << im.cols; - } - this->transform(im, trg); -} - -int DataTransformer::Rand(int min, int max) { - std::random_device source; - std::mt19937 rng(source()); - std::uniform_int_distribution dist(min, max); - return dist(rng); -} - -void DataTransformer::transform(cv::Mat& cvImgOri, float* target) { - const int imgChannels = cvImgOri.channels(); - const int imgHeight = cvImgOri.rows; - const int imgWidth = cvImgOri.cols; - const bool doMirror = (!isTest_) && Rand(0, 1); - int h_off = 0; - int w_off = 0; - int th = imgHeight; - int tw = imgWidth; - cv::Mat img; - if (imgSize_ > 0) { - if (imgHeight > imgWidth) { - tw = imgSize_; - th = int(double(imgHeight) / imgWidth * tw); - th = th > imgSize_ ? th : imgSize_; - } else { - th = imgSize_; - tw = int(double(imgWidth) / imgHeight * th); - tw = tw > imgSize_ ? tw : imgSize_; - } - cv::resize(cvImgOri, img, cv::Size(tw, th)); - } else { - cv::Mat img = cvImgOri; - } - - cv::Mat cv_cropped_img = img; - if (cropHeight_ && cropWidth_) { - if (!isTest_) { - h_off = Rand(0, th - cropHeight_); - w_off = Rand(0, tw - cropWidth_); - } else { - h_off = (th - cropHeight_) / 2; - w_off = (tw - cropWidth_) / 2; - } - cv::Rect roi(w_off, h_off, cropWidth_, cropHeight_); - cv_cropped_img = img(roi); - } else { - CHECK_EQ(cropHeight_, imgHeight); - CHECK_EQ(cropWidth_, imgWidth); - } - int height = cropHeight_; - int width = cropWidth_; - int top_index; - for (int h = 0; h < height; ++h) { - const uchar* ptr = cv_cropped_img.ptr(h); - int img_index = 0; - for (int w = 0; w < width; ++w) { - for (int c = 0; c < imgChannels; ++c) { - if (doMirror) { - top_index = (c * height + h) * width + width - 1 - w; - } else { - top_index = (c * height + h) * width + w; - } - float pixel = static_cast(ptr[img_index++]); - if (isEltMean_) { - int mean_index = (c * imgHeight + h) * imgWidth + w; - target[top_index] = (pixel - meanValues_[mean_index]) * scale_; - } else { - if (isChannelMean_) { - target[top_index] = (pixel - meanValues_[c]) * scale_; - } else { - target[top_index] = pixel * scale_; - } - } - } - } - } // target: BGR -} - -void DataTransformer::start(std::vector& data, - int* datalen, - int* labels) { - auto job = [&](int tid, int numThreads) { - for (size_t i = tid; i < data.size(); i += numThreads) { - DataTypePtr ret = prefetchFree_.dequeue(); - char* buf = data[i]; - int size = datalen[i]; - ret->second = labels[i]; - this->startFetching(buf, size, ret->first); - prefetchFull_.enqueue(ret); - } - }; - syncThreadPool_->exec(job); - fetchCount_ = data.size(); -} - -void DataTransformer::obtain(float* data, int* label) { - fetchCount_--; - if (fetchCount_ < 0) { - LOG(FATAL) << "Empty data"; - } - DataTypePtr ret = prefetchFull_.dequeue(); - *label = ret->second; - memcpy(data, ret->first, sizeof(float) * imgPixels_); - prefetchFree_.enqueue(ret); -} diff --git a/plugin/opencv/DataTransformer.h b/plugin/opencv/DataTransformer.h deleted file mode 100644 index 603cea3059..0000000000 --- a/plugin/opencv/DataTransformer.h +++ /dev/null @@ -1,122 +0,0 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. - -Licensed under the Apache License, Version 2.0 (the "License"); -you may not use this file except in compliance with the License. -You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - -Unless required by applicable law or agreed to in writing, software -distributed under the License is distributed on an "AS IS" BASIS, -WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -See the License for the specific language governing permissions and -limitations under the License. */ - -#ifndef DATATRANSFORMER_H_ -#define DATATRANSFORMER_H_ - -#include -#include -#include -#include -#include -#include - -#include "paddle/utils/Thread.h" - -/** - * This is an image processing module with OpenCV, such as - * resizing, scaling, mirroring, substracting the image mean... - * - * This class has a double BlockQueue and they shared the same memory. - * It is used to avoid create memory each time. And it also can - * return the data even if the data are processing in multi-threads. - */ -class DataTransformer { -public: - DataTransformer(int threadNum, - int capacity, - bool isTest, - bool isColor, - int cropHeight, - int cropWidth, - int imgSize, - bool isEltMean, - bool isChannelMean, - float* meanValues); - virtual ~DataTransformer() { - if (meanValues_) { - free(meanValues_); - } - } - - /** - * @brief Start multi-threads to transform a list of input data. - * The processed data will be saved in Queue of prefetchFull_. - * - * @param data Data containing the image string to be transformed. - * @param label The label of input image. - */ - void start(std::vector& data, int* datalen, int* labels); - - /** - * @brief Applies the transformation on one image Mat. - * - * @param img The input img to be transformed. - * @param target target is used to save the transformed data. - */ - void transform(cv::Mat& img, float* target); - - /** - * @brief Decode the image string, then calls transform() function. - * - * @param src The input image string. - * @param size The length of string. - * @param trg trg is used to save the transformed data. - */ - void startFetching(const char* src, const int size, float* trg); - - /** - * @brief Return the transformed data and its label. - */ - void obtain(float* data, int* label); - -private: - int isTest_; - int isColor_; - int cropHeight_; - int cropWidth_; - int imgSize_; - int capacity_; - int fetchCount_; - bool isEltMean_; - bool isChannelMean_; - int numThreads_; - float scale_; - int imgPixels_; - float* meanValues_; - - /** - * Initialize the mean values. - */ - void loadMean(float* values); - - /** - * @brief Generates a random integer from Uniform({min, min + 1, ..., max}). - * @param min The lower bound (inclusive) value of the random number. - * @param max The upper bound (inclusive) value of the random number. - * - * @return - * A uniformly random integer value from ({min, min + 1, ..., max}). - */ - int Rand(int min, int max); - - typedef std::pair DataType; - typedef std::shared_ptr DataTypePtr; - std::vector prefetch_; - std::unique_ptr syncThreadPool_; - paddle::BlockingQueue prefetchFree_; - paddle::BlockingQueue prefetchFull_; -}; // class DataTransformer - -#endif // DATATRANSFORMER_H_ diff --git a/plugin/opencv/PyDecodejpeg.cpp b/plugin/opencv/PyDecodejpeg.cpp deleted file mode 100644 index a32e6430e1..0000000000 --- a/plugin/opencv/PyDecodejpeg.cpp +++ /dev/null @@ -1,175 +0,0 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. - -Licensed under the Apache License, Version 2.0 (the "License"); -you may not use this file except in compliance with the License. -You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - -Unless required by applicable law or agreed to in writing, software -distributed under the License is distributed on an "AS IS" BASIS, -WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -See the License for the specific language governing permissions and -limitations under the License. */ - -#include -#include -#include -#include -#include -#include -#include -#include - -#include "DataTransformer.h" - -/** - * DecodeJpeg is an image processing API for interfacing Python and C++ - * code DataTransformer, which used OpenCV and multi-threads to accelerate - * image processing. - * The Boost Python Library is used to wrap C++ interfaces. - */ - -class DecodeJpeg { -public: - /** - * The constructor will create and initialize an object of DataTransformer. - */ - DecodeJpeg(int threadNum, - int capacity, - bool isTest, - bool isColor, - int resize_min_size, - int cropSizeH, - int cropSizeW, - PyObject* meanValues) { - int channel = isColor ? 3 : 1; - bool isEltMean = false; - bool isChannelMean = false; - float* mean = NULL; - if (meanValues || meanValues != Py_None) { - if (!PyArray_Check(meanValues)) { - LOG(FATAL) << "Object is not a numpy array"; - } - pyTypeCheck(meanValues); - int size = PyArray_SIZE(reinterpret_cast(meanValues)); - isChannelMean = (size == channel) ? true : false; - isEltMean = (size == channel * cropSizeH * cropSizeW) ? true : false; - CHECK(isChannelMean != isEltMean); - mean = (float*)PyArray_DATA(reinterpret_cast(meanValues)); - } - tfhandlerPtr_ = std::make_shared(threadNum, - capacity, - isTest, - isColor, - cropSizeH, - cropSizeW, - resize_min_size, - isEltMean, - isChannelMean, - mean); - } - - ~DecodeJpeg() {} - - /** - * @brief This function is used to parse the Python object and convert - * the data to C++ format. Then it called the function of - * DataTransformer to start image processing. - * @param pysrc The input image list with string type. - * @param pylabel The input label of image. - * It's type is numpy.array with int32. - */ - void start(boost::python::list& pysrc, PyObject* pydlen, PyObject* pylabel) { - std::vector data; - int num = len(pysrc); - for (int t = 0; t < num; ++t) { - char* src = boost::python::extract(pysrc[t]); - data.push_back(src); - } - int* dlen = (int*)PyArray_DATA(reinterpret_cast(pydlen)); - int* dlabels = - (int*)PyArray_DATA(reinterpret_cast(pylabel)); - tfhandlerPtr_->start(data, dlen, dlabels); - } - - /** - * @brief Return one processed data. - * @param pytrg The processed image. - * @param pylabel The label of processed image. - */ - void get(PyObject* pytrg, PyObject* pylab) { - pyWritableCheck(pytrg); - pyWritableCheck(pylab); - pyContinuousCheck(pytrg); - pyContinuousCheck(pylab); - float* data = (float*)PyArray_DATA(reinterpret_cast(pytrg)); - int* label = (int*)PyArray_DATA(reinterpret_cast(pylab)); - tfhandlerPtr_->obtain(data, label); - } - - /** - * @brief An object of DataTransformer, which is used to call - * the image processing funtions. - */ - std::shared_ptr tfhandlerPtr_; - -private: - /** - * @brief Check whether the type of PyObject is valid or not. - */ - void pyTypeCheck(PyObject* o) { - int typenum = PyArray_TYPE(reinterpret_cast(o)); - - // clang-format off - int type = - typenum == NPY_UBYTE ? CV_8U : - typenum == NPY_BYTE ? CV_8S : - typenum == NPY_USHORT ? CV_16U : - typenum == NPY_SHORT ? CV_16S : - typenum == NPY_INT || typenum == NPY_LONG ? CV_32S : - typenum == NPY_FLOAT ? CV_32F : - typenum == NPY_DOUBLE ? CV_64F : -1; - // clang-format on - - if (type < 0) { - LOG(FATAL) << "toMat: Data type = " << type << " is not supported"; - } - } - - /** - * @brief Check whether the PyObject is writable or not. - */ - void pyWritableCheck(PyObject* o) { - CHECK(PyArray_ISWRITEABLE(reinterpret_cast(o))); - } - - /** - * @brief Check whether the PyObject is c-contiguous or not. - */ - void pyContinuousCheck(PyObject* o) { - CHECK(PyArray_IS_C_CONTIGUOUS(reinterpret_cast(o))); - } -}; // DecodeJpeg - -/** - * @brief Initialize the Python interpreter and numpy. - */ -static void initPython() { - Py_Initialize(); - PyOS_sighandler_t sighandler = PyOS_getsig(SIGINT); - import_array(); - PyOS_setsig(SIGINT, sighandler); -} - -/** - * Use Boost.Python to expose C++ interface to Python. - */ -BOOST_PYTHON_MODULE(DeJpeg) { - initPython(); - boost::python::class_( - "DecodeJpeg", - boost::python::init()) - .def("start", &DecodeJpeg::start) - .def("get", &DecodeJpeg::get); -}; diff --git a/python/paddle/utils/image_multiproc.py b/python/paddle/utils/image_multiproc.py index ccc0a531a7..82df6d6c0c 100644 --- a/python/paddle/utils/image_multiproc.py +++ b/python/paddle/utils/image_multiproc.py @@ -1,44 +1,50 @@ -import os, psutil -import cv2 -from paddle.utils.image_util import * +import os, sys +import numpy as np +from PIL import Image +from cStringIO import StringIO import multiprocessing -import subprocess, signal, sys +from functools import partial + +from paddle.utils.image_util import * +from paddle.trainer.config_parser import logger +try: + import cv2 +except ImportError: + logger.warning("OpenCV2 is not installed, using PIL to prcoess") + cv2 = None -class CvImageTransfomer(ImageTransformer): + +class CvTransfomer(ImageTransformer): """ - CvImageTransfomer used python-opencv to process image. + CvTransfomer used python-opencv to process image. """ - def __init__(self, - min_size=None, - crop_size=None, - transpose=None, - channel_swap=None, - mean=None, - is_train=True, - is_color=True): + def __init__( + self, + min_size=None, + crop_size=None, + transpose=(2, 0, 1), # transpose to C * H * W + channel_swap=None, + mean=None, + is_train=True, + is_color=True): ImageTransformer.__init__(self, transpose, channel_swap, mean, is_color) self.min_size = min_size self.crop_size = crop_size self.is_train = is_train - def cv_resize_fixed_short_side(self, im, min_size): + def resize(self, im, min_size): row, col = im.shape[:2] - scale = min_size / float(min(row, col)) - if row < col: - row = min_size - col = int(round(col * scale)) - col = col if col > min_size else min_size + new_row, new_col = min_size, min_size + if row > col: + new_row = min_size * row / col else: - col = min_size - row = int(round(row * scale)) - row = row if row > min_size else min_size - resized_size = row, col - im = cv2.resize(im, resized_size, interpolation=cv2.INTER_CUBIC) + new_col = min_size * col / row + im = cv2.resize(im, (new_row, new_col), interpolation=cv2.INTER_CUBIC) return im - def crop_img(self, im): + def crop_and_flip(self, im): """ Return cropped image. The size of the cropped image is inner_size * inner_size. @@ -65,8 +71,8 @@ class CvImageTransfomer(ImageTransformer): return im def transform(self, im): - im = self.cv_resize_fixed_short_side(im, self.min_size) - im = self.crop_img(im) + im = self.resize(im, self.min_size) + im = self.crop_and_flip(im) # transpose, swap channel, sub mean im = im.astype('float32') ImageTransformer.transformer(self, im) @@ -81,90 +87,187 @@ class CvImageTransfomer(ImageTransformer): im = self.load_image_from_string(data) return self.transform(im) + def load_image_from_file(self, file): + flag = cv2.CV_LOAD_IMAGE_COLOR if self.is_color else cv2.CV_LOAD_IMAGE_GRAYSCALE + im = cv2.imread(file, flag) + return im + + def transform_from_file(self, file): + im = self.load_image_from_file(file) + return self.transform(im) + + +class PILTransfomer(ImageTransformer): + """ + PILTransfomer used PIL to process image. + """ + + def __init__( + self, + min_size=None, + crop_size=None, + transpose=(2, 0, 1), # transpose to C * H * W + channel_swap=None, + mean=None, + is_train=True, + is_color=True): + ImageTransformer.__init__(self, transpose, channel_swap, mean, is_color) + self.min_size = min_size + self.crop_size = crop_size + self.is_train = is_train + + def resize(self, im, min_size): + row, col = im.size[:2] + new_row, new_col = min_size, min_size + if row > col: + new_row = min_size * row / col + else: + new_col = min_size * col / row + im = im.resize((new_row, new_col), Image.ANTIALIAS) + return im -class MultiProcessImageTransfomer(): + def crop_and_flip(self, im): + """ + Return cropped image. + The size of the cropped image is inner_size * inner_size. + """ + row, col = im.size[:2] + start_h, start_w = 0, 0 + if self.is_train: + start_h = np.random.randint(0, row - self.crop_size + 1) + start_w = np.random.randint(0, col - self.crop_size + 1) + else: + start_h = (row - self.crop_size) / 2 + start_w = (col - self.crop_size) / 2 + end_h, end_w = start_h + self.crop_size, start_w + self.crop_size + im = im.crop((start_h, start_w, end_h, end_w)) + if (self.is_train) and (np.random.randint(2) == 0): + im = im.transpose(Image.FLIP_LEFT_RIGHT) + return im + + def transform(self, im): + im = self.resize(im, self.min_size) + im = self.crop_and_flip(im) + im = np.array(im, dtype=np.float32) # convert to numpy.array + # transpose, swap channel, sub mean + ImageTransformer.transformer(self, im) + return im + + def load_image_from_string(self, data): + im = Image.open(StringIO(data)) + return im + + def transform_from_string(self, data): + im = self.load_image_from_string(data) + return self.transform(im) + + def load_image_from_file(self, file): + im = Image.open(file) + return im + + def transform_from_file(self, file): + im = self.load_image_from_file(file) + return self.transform(im) + + +def warpper(cls, (dat, label)): + return cls.job(dat, label) + + +class MultiProcessImageTransformer(object): def __init__(self, procnum=10, - capacity=10240, - min_size=None, + resize_size=None, crop_size=None, - transpose=None, + transpose=(2, 0, 1), channel_swap=None, mean=None, is_train=True, - is_color=True): - self.procnum = procnum - self.capacity = capacity - self.size = 0 - self.count = 0 - signal.signal(signal.SIGTERM, self.kill_child_processes) - self.fetch_queue = multiprocessing.Queue(maxsize=capacity) - self.cv_transformer = CvImageTransfomer(min_size, crop_size, transpose, - channel_swap, mean, is_train, - is_color) - - def __del__(self): - try: - for p in self.procs: - p.join() - except Exception as e: - print str(e) - - def reset(self, size): - self.size = size - self.count = 0 - self.procs = [] - - def run_proc(self, data, label): - dlen = len(label) - self.reset(dlen) - for i in xrange(self.procnum): - start = dlen * i / self.procnum - end = dlen * (i + 1) / self.procnum - proc = multiprocessing.Process( - target=self.batch_transfomer, - args=(data[start:end], label[start:end])) - proc.daemon = True - self.procs.append(proc) - for p in self.procs: - p.start() - - def get(self): - """ - Return one processed image. - """ - # block if necessary until an item is available - data, lab = self.fetch_queue.get(block=True) - self.count += 1 - if self.count == self.size: - try: - for p in self.procs: - p.join() - except Exception as e: - print str(e) - return data, lab - - def batch_transfomer(self, data, label): + is_color=True, + is_img_string=True): """ - param data: input data in format of image string - type data: a list of string - label: the label of image - """ - for i in xrange(len(label)): - res = self.cv_transformer.transform_from_string(data[i]) - self.fetch_queue.put((res, int(label[i]))) + Processing image with multi-process. If it is used in PyDataProvider, + the simple usage for CNN is as follows: + + .. code-block:: python - def kill_child_processes(self, signum, frame): - """ - Kill a process's child processes in python. + def hool(settings, is_train, **kwargs): + settings.is_train = is_train + settings.mean_value = np.array([103.939,116.779,123.68], dtype=np.float32) + settings.input_types = [ + dense_vector(3 * 224 * 224), + integer_value(1)] + settings.transformer = MultiProcessImageTransformer( + procnum=10, + resize_size=256, + crop_size=224, + transpose=(2, 0, 1), + mean=settings.mean_values, + is_train=settings.is_train) + + + @provider(init_hook=hook, pool_size=20480) + def process(settings, file_list): + with open(file_list, 'r') as fdata: + for line in fdata: + data_dic = np.load(line.strip()) # load the data batch pickled by Pickle. + data = data_dic['data'] + labels = data_dic['label'] + labels = np.array(labels, dtype=np.float32) + for im, lab in settings.dp.run(data, labels): + yield [im.astype('float32'), int(lab)] + + :param procnum: processor number. + :type procnum: int + :param resize_size: the shorter edge size of image after resizing. + :type resize_size: int + :param crop_size: the croping size. + :type crop_size: int + :param transpose: the transpose order, Paddle only allow C * H * W order. + :type transpose: tuple or list + :param channel_swap: the channel swap order, RGB or BRG. + :type channel_swap: tuple or list + :param mean: the mean values of image, per-channel mean or element-wise mean. + :type mean: array, The dimension is 1 for per-channel mean. + The dimension is 3 for element-wise mean. + :param is_train: training peroid or testing peroid. + :type is_train: bool. + :param is_color: the image is color or gray. + :type is_color: bool. + :param is_img_string: The input can be the file name of image or image string. + :type is_img_string: bool. """ - parent_id = os.getpid() - ps_command = subprocess.Popen( - "ps -o pid --ppid %d --noheaders" % parent_id, - shell=True, - stdout=subprocess.PIPE) - ps_output = ps_command.stdout.read() - retcode = ps_command.wait() - for pid_str in ps_output.strip().split("\n")[:-1]: - os.kill(int(pid_str), signal.SIGTERM) - sys.exit() + + self.pool = multiprocessing.Pool(procnum) + self.is_img_string = is_img_string + if cv2 is not None: + self.transformer = CvTransfomer(resize_size, crop_size, transpose, + channel_swap, mean, is_train, + is_color) + else: + self.transformer = PILTransfomer(resize_size, crop_size, transpose, + channel_swap, mean, is_train, + is_color) + + def run(self, data, label): + try: + fun = partial(warpper, self) + return self.pool.imap_unordered(fun, zip(data, label), chunksize=5) + except KeyboardInterrupt: + self.pool.terminate() + except Exception, e: + self.pool.terminate() + + def job(self, data, label): + if self.is_img_string: + return self.transformer.transform_from_string(data), label + else: + return self.transformer.transform_from_file(data), label + + def __getstate__(self): + self_dict = self.__dict__.copy() + del self_dict['pool'] + return self_dict + + def __setstate__(self, state): + self.__dict__.update(state) From 7f78912c9fa4e5b9b265738584a75e89a63513c1 Mon Sep 17 00:00:00 2001 From: hanchao Date: Mon, 5 Dec 2016 21:07:31 +0800 Subject: [PATCH 19/60] test code for issue #729 --- .../tests/configs/generate_protostr.sh | 2 + .../tests/configs/run_tests.sh | 11 +++-- .../test_config_parser_for_non_file_config.py | 40 +++++++++++++++++++ 3 files changed, 49 insertions(+), 4 deletions(-) create mode 100644 python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py diff --git a/python/paddle/trainer_config_helpers/tests/configs/generate_protostr.sh b/python/paddle/trainer_config_helpers/tests/configs/generate_protostr.sh index e55f9bd388..a54af94ce3 100755 --- a/python/paddle/trainer_config_helpers/tests/configs/generate_protostr.sh +++ b/python/paddle/trainer_config_helpers/tests/configs/generate_protostr.sh @@ -11,10 +11,12 @@ for conf in ${configs[*]} do echo "Generating " $conf python -m paddle.utils.dump_config $conf.py > $protostr/$conf.protostr.unittest + cat ${conf}.py |python test_config_parser_for_non_file_config.py > $protostr/$conf.protostr.non_file_config.unittest done for conf in ${whole_configs[*]} do echo "Generating " $conf python -m paddle.utils.dump_config $conf.py "" --whole > $protostr/$conf.protostr.unittest + cat ${conf}.py |python test_config_parser_for_non_file_config.py --whole > $protostr/$conf.protostr.non_file_config.unittest done diff --git a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh index 73f8b333b2..ed2ac6ed18 100755 --- a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh +++ b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh @@ -16,20 +16,23 @@ if [ -z $1 ]; then do base_protostr=$protostr/$file new_protostr=$protostr/$file.unittest - diff $base_protostr $new_protostr -u + diff $base_protostr $new_protostr -u && + diff $protostr/$file $protostr/$file.non_file_config.unittest -u done else for file in ${configs[*]} do if ! $1 $protostr/$file.protostr $protostr/$file.protostr.unittest; then - diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u + diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u && + diff $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest -u fi done for file in ${whole_configs[*]} - do +do if ! $1 $protostr/$file.protostr $protostr/$file.protostr.unittest --whole; then - diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u + diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u && + diff $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest -u fi done fi diff --git a/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py new file mode 100644 index 0000000000..71ee0499d1 --- /dev/null +++ b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py @@ -0,0 +1,40 @@ +#!/usr/bin/env python +# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import sys +import getopt + +whole = False +opts, args = getopt.getopt(sys.argv[1:], "", ["whole"]) +for op, value in opts: + if op == "--whole": + whole = True + +cmdstr = """ +from paddle.trainer.config_parser import * +from paddle.trainer_config_helpers import * +def configs():\n""" + +for line in sys.stdin: + if "import" in line and "from" in line: + continue + cmdstr = cmdstr + " " + line + +if whole: + cmdstr = cmdstr + """print parse_config(configs, "")""" +else: + cmdstr = cmdstr + """print parse_config(configs, "").model_config""" + +exec(cmdstr) From 84d47ac205b81ff06efd73892cd714f7874dda63 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Tue, 6 Dec 2016 11:43:14 +0800 Subject: [PATCH 20/60] follow comments --- python/paddle/utils/image_multiproc.py | 55 +++++++++++--------------- 1 file changed, 22 insertions(+), 33 deletions(-) diff --git a/python/paddle/utils/image_multiproc.py b/python/paddle/utils/image_multiproc.py index 82df6d6c0c..6ce32f7811 100644 --- a/python/paddle/utils/image_multiproc.py +++ b/python/paddle/utils/image_multiproc.py @@ -3,7 +3,8 @@ import numpy as np from PIL import Image from cStringIO import StringIO import multiprocessing -from functools import partial +import functools +import itertools from paddle.utils.image_util import * from paddle.trainer.config_parser import logger @@ -14,10 +15,12 @@ except ImportError: logger.warning("OpenCV2 is not installed, using PIL to prcoess") cv2 = None +__all__ = ["CvTransformer", "PILTransformer", "MultiProcessImageTransformer"] -class CvTransfomer(ImageTransformer): + +class CvTransformer(ImageTransformer): """ - CvTransfomer used python-opencv to process image. + CvTransformer used python-opencv to process image. """ def __init__( @@ -97,9 +100,9 @@ class CvTransfomer(ImageTransformer): return self.transform(im) -class PILTransfomer(ImageTransformer): +class PILTransformer(ImageTransformer): """ - PILTransfomer used PIL to process image. + PILTransformer used PIL to process image. """ def __init__( @@ -170,8 +173,11 @@ class PILTransfomer(ImageTransformer): return self.transform(im) -def warpper(cls, (dat, label)): - return cls.job(dat, label) +def job(is_img_string, transformer, (data, label)): + if is_img_string: + return transformer.transform_from_string(data), label + else: + return transformer.transform_from_file(data), label class MultiProcessImageTransformer(object): @@ -238,36 +244,19 @@ class MultiProcessImageTransformer(object): :type is_img_string: bool. """ + self.procnum = procnum self.pool = multiprocessing.Pool(procnum) self.is_img_string = is_img_string if cv2 is not None: - self.transformer = CvTransfomer(resize_size, crop_size, transpose, - channel_swap, mean, is_train, - is_color) - else: - self.transformer = PILTransfomer(resize_size, crop_size, transpose, + self.transformer = CvTransformer(resize_size, crop_size, transpose, channel_swap, mean, is_train, is_color) - - def run(self, data, label): - try: - fun = partial(warpper, self) - return self.pool.imap_unordered(fun, zip(data, label), chunksize=5) - except KeyboardInterrupt: - self.pool.terminate() - except Exception, e: - self.pool.terminate() - - def job(self, data, label): - if self.is_img_string: - return self.transformer.transform_from_string(data), label else: - return self.transformer.transform_from_file(data), label - - def __getstate__(self): - self_dict = self.__dict__.copy() - del self_dict['pool'] - return self_dict + self.transformer = PILTransformer(resize_size, crop_size, transpose, + channel_swap, mean, is_train, + is_color) - def __setstate__(self, state): - self.__dict__.update(state) + def run(self, data, label): + fun = functools.partial(job, self.is_img_string, self.transformer) + return self.pool.imap_unordered( + fun, itertools.izip(data, label), chunksize=100 * self.procnum) From 0592d1b76eb3ac9bc3c6e3a479cb72dfab1d45e8 Mon Sep 17 00:00:00 2001 From: Li Peng Date: Wed, 7 Dec 2016 18:10:17 +0800 Subject: [PATCH 21/60] Add bazel installation to the Dockerfiles Also add DEBIAN_FRONTEND=noninteractive to suppress the debconf messages. --- paddle/scripts/docker/Dockerfile | 8 ++++++++ paddle/scripts/docker/Dockerfile.gpu | 8 ++++++++ 2 files changed, 16 insertions(+) diff --git a/paddle/scripts/docker/Dockerfile b/paddle/scripts/docker/Dockerfile index 2a1a842336..a9a72b355e 100644 --- a/paddle/scripts/docker/Dockerfile +++ b/paddle/scripts/docker/Dockerfile @@ -12,6 +12,14 @@ RUN apt-get update \ RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark +ARG DEBIAN_FRONTEND=noninteractive +RUN apt-get update && apt-get install -y curl software-properties-common \ + && add-apt-repository ppa:webupd8team/java \ + && echo "oracle-java8-installer shared/accepted-oracle-license-v1-1 select true" | debconf-set-selections \ + && echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | tee /etc/apt/sources.list.d/bazel.list \ + && curl https://bazel.build/bazel-release.pub.gpg | apt-key add - \ + && apt-get update && apt-get install -y oracle-java8-installer bazel + ARG WITH_AVX ARG WITH_DOC ARG WITH_SWIG_PY diff --git a/paddle/scripts/docker/Dockerfile.gpu b/paddle/scripts/docker/Dockerfile.gpu index b3253d23c3..a147e3840e 100644 --- a/paddle/scripts/docker/Dockerfile.gpu +++ b/paddle/scripts/docker/Dockerfile.gpu @@ -12,6 +12,14 @@ RUN apt-get update \ RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark +ARG DEBIAN_FRONTEND=noninteractive +RUN apt-get update && apt-get install -y curl software-properties-common \ + && add-apt-repository ppa:webupd8team/java \ + && echo "oracle-java8-installer shared/accepted-oracle-license-v1-1 select true" | debconf-set-selections \ + && echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | tee /etc/apt/sources.list.d/bazel.list \ + && curl https://bazel.build/bazel-release.pub.gpg | apt-key add - \ + && apt-get update && apt-get install -y oracle-java8-installer bazel + ARG WITH_AVX ARG WITH_DOC ARG WITH_SWIG_PY From 00b86b1f3a967357517bd0acace0e166273afb82 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Wed, 7 Dec 2016 21:37:19 +0800 Subject: [PATCH 22/60] Follow comments --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 58 +++++++++++-------- 1 file changed, 35 insertions(+), 23 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 7ae9f5ef8e..3f5100cbf1 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -4,14 +4,14 @@ 单双层RNN API对比介绍 ##################### -这篇教程主要介绍了\ :ref:`glossary_双层RNN`\ 的API接口。本文以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。本文中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 +本文以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。本文中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 示例1：双层RNN，子序列间无Memory ================================ -在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_sequence`\ 数据，分别进行序列操作。并且内层的序列操作之间是独立没有依赖的，即不需要使用\ :ref:`glossary_Memory`\ 的。 +在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_sequence`\ 数据，分别进行序列操作；并且内层的序列操作之间独立无依赖，即不需要使用\ :ref:`glossary_Memory`\ 。 -在本示例中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体，同时使用encoder压缩，二者语意上完全一致。这组语意相同的示例配置如下 +在本示例中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体同时使用encoder压缩。二者语意上完全一致。这组语义相同的示例配置如下： * 单层\ :ref:`glossary_RNN`\: `sequence_layer_group.conf `_ * :ref:`glossary_双层RNN`\: `sequence_nest_layer_group.conf `_ @@ -28,7 +28,7 @@ :language: text -- 双层序列数据一共有4个样本。 每个样本间用空行分开，整体数据和原始数据完全一样。而对于双层序列的LSTM来说，第一条数据同时encode两条数据成两个向量。这四条数据同时处理的句子为\ :code:`[2, 3, 2, 3]`\ 。 +- 双层序列数据一共有4个样本。 每个样本间用空行分开，整体数据和原始数据完全一样。但于双层序列的LSTM来说，第一个样本同时encode两条数据成两个向量。这四条数据同时处理的句子数量为\ :code:`[2, 3, 2, 3]`\ 。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest :language: text @@ -51,17 +51,17 @@ :lines: 42-71 :linenos: -- 这是对于同样的数据，本示例中双层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： +- 对于同样的数据，双层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 的代码。其说明如下： - :ref:`glossary_DataProvider`\ 共返回两组数据，分别是sentences和labels。即在双层序列的原始数据中，每一组内的所有句子和labels - - sentences是双层\ :ref:`glossary_sequence`\ 的数据。他内部包括了每组数据中的所有句子，又使用句子中每一个单词的词表index表示每一个句子，故为双层\ :ref:`glossary_sequence`\ 。类型为 integer_value_sub_sequence 。 - - labels是每组内每一个句子的标签，故而是一个单层\ :ref:`glossary_sequence`\ 。 + - sentences是双层\ :ref:`glossary_sequence`\ 的数据。由于它内部包含了每组数据中的所有句子，且每个句子表示为对应的词表索引数组，因此它是integer_value_sub_sequence 类型的，即双层\ :ref:`glossary_sequence`\ 。 + - labels是每组内每个句子的标签，故而是一个单层\ :ref:`glossary_sequence`\ 。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个LSTM网络。 +首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行(高亮部分)即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个LSTM网络。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf :language: python @@ -70,19 +70,19 @@ :emphasize-lines: 9-15 -其次，我们看一下语义相同的\ :ref:`glossary_双层RNN`\ 的网络配置。 +其次，我们看一下语义相同的\ :ref:`glossary_双层RNN`\ 的网络配置\: * PaddlePaddle中的许多layer并不在意输入是否是\ :ref:`glossary_sequence`\ ，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个\ :ref:`glossary_timestep`\ 来进行的。 -* 在该配置中，7-26行将双层\ :ref:`glossary_sequence`\ 数据，先变换成单层\ :ref:`glossary_sequence`\ 数据，在对每一个单层\ :ref:`glossary_sequence`\ 进行处理。 +* 在该配置的7-26行(高亮部分)，将双层\ :ref:`glossary_sequence`\ 数据先变换成单层\ :ref:`glossary_sequence`\ 数据，再对每一个单层\ :ref:`glossary_sequence`\ 进行处理。 * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层\ :ref:`glossary_sequence`\ => 单层\ :ref:`glossary_sequence`\ ，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层\ :ref:`glossary_RNN`\ 的配置是等价的。 -* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要知道使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但是，和单层\ :ref:`glossary_RNN`\ 有区别的地方是，我们是对每一个子序列取最后一个元素。于是我们设置\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 +* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但和单层\ :ref:`glossary_RNN`\ 不同，我们是对每一个子序列取最后一个元素，因此\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 -* 至此，\ :code:`lstm_last`\ 便和单层\ :ref:`glossary_RNN`\ 的配置中的\ :code:`lstm_last`\ 具有相同的结果了。 +* 至此，\ :code:`lstm_last`\ 便和单层\ :ref:`glossary_RNN`\ 配置中的\ :code:`lstm_last`\ 具有相同的结果了。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf :language: python @@ -93,30 +93,34 @@ 示例2：:ref:`glossary_双层RNN`，子序列间有\ :ref:`glossary_Memory` ================================================================== -本示例中，意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 同时实现一个完全等价的全连接\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。而对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 +本示例意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 + +* 对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。 + +* 对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ 我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 -- 单层序列：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 +- 单层\ :ref:`glossary_rnn`\ ：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf :language: python :lines: 36-48 -- 双层序列，外层memory是一个元素： +- \ :ref:`glossary_双层RNN`\ ，外层memory是一个元素： - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每一个时间步都用了上一个时间步的输出结果”一致。 + - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每个时间步都用了上一个时间步的输出结果”一致。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn.conf :language: python :lines: 39-66 .. warning:: - PaddlePaddle目前只支持在每一个时间步中，Memory的sequence长度一致的情况。 + PaddlePaddle目前只支持在每个时间步中，Memory的\ :ref:`glossary_sequence`\ 长度一致的情况。 示例3：双层RNN，输入不等长 ========================== @@ -127,31 +131,39 @@ -**输入不等长** 是指recurrent_group的多个输入序列，在每个\ :ref:`glossary_timestep`\ 的子序列长度可以不相等。但\ :ref:`glossary_双层RNN`\ 目前需要整体的输出，与某一个输入的序列信息是一致的。使用\ :red:`targetInlink`\ 可以指定和输出序列信息一致。 +**输入不等长** 是指recurrent_group的多个输入序列，在每个\ :ref:`glossary_timestep`\ 的子序列长度可以不相等。但序列输出时，需要指定与某一个输入的序列信息是一致的。使用\ :red:`targetInlink`\ 可以指定哪一个输入和输出序列信息一致，默认指定第一个输入。 + +示例3的配置分别为\ `单层不等长RNN `_\ 和\ `双层不等长RNN `_\ 。 -本例参考配置分别为\ `单层不等长RNN `_\ 和\ `双层不等长RNN `_\ 。 +示例3对于单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 数据完全相同。 -本例中对于单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 数据完全相同，对于单层\ :ref:`glossary_RNN`\ 的数据一共有两个样本，他们分别是\ :code:`[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]`\ 和\ :code:`[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]`\ 。对于每一个单层\ :ref:`glossary_RNN`\ 的数据，均有两组特征。在单层数据的基础上，\ :ref:`glossary_双层RNN`\ 数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。但是需要注意的是Paddle目前只支持序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。即两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 +* 对于单层\ :ref:`glossary_RNN`\ 的数据一共有两个样本，他们分别是\ :code:`[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]`\ 和\ :code:`[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]`\ 。对于每一个单层\ :ref:`glossary_RNN`\ 的数据，均有两组特征。 + +* 在单层数据的基础上，\ :ref:`glossary_双层RNN`\ 数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。 + +* 需要注意的是Paddle目前只支持子序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。例如本里中的两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -本例中的配置，使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 使用一个\ :code:`recurrent_group`\ 将两个序列同时过完全连接的\ :ref:`glossary_RNN`\ 。对于单层\ :ref:`glossary_RNN`\ 的code如下。 +和示例2中的配置累死，示例3的配置使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ ，实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 + +* 单层\ :ref:`glossary_RNN`\ \: .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py :language: python :lines: 42-59 :linenos: -而双层序列的代码如下。 +* :ref:`glossary_双层RNN`\ \: .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py :language: python :lines: 41-80 :linenos: -在上面代码中，单层和双层序列的使用和示例2中的示例类似，区别是同时处理了两个输入。而对于双层序列，两个输入的子序列长度也并不相同。但是，我们使用了\ :code:`targetInlink`\ 参数设置了外层\ :code:`recurrent_group`\ 的输出格式。所以外层输出的序列形状，和\ :code:`emb2`的序列形状一致。 +在上面代码中，单层和双层序列的使用和示例2中的示例类似，区别是同时处理了两个输入。而对于双层序列，两个输入的子序列长度也并不相同。但是，我们使用了\ :code:`targetInlink`\ 参数设置了外层\ :code:`recurrent_group`\ 的输出格式。所以外层输出的序列形状，和\ :code:`emb2`\ 的序列形状一致。 示例4：beam_search的生成 ======================== From 3cf7337f3055f8f6aec15365a0fe2756acc5cf48 Mon Sep 17 00:00:00 2001 From: xuwei06 Date: Wed, 7 Dec 2016 13:51:20 -0800 Subject: [PATCH 23/60] Correctly handling multiple calls to parse_config() To solve this, we maintain the list of DefaultNameFactory used in by trainer_config_helper, and reset the state at the beginning of each parse_config call. Change-Id: I13c7574dc8f0b6bc6f6b7c92eb425e2c52c926e8 --- python/paddle/trainer/config_parser.py | 18 +++++++++--- .../default_decorators.py | 15 +++++++++- .../tests/CMakeLists.txt | 5 ++++ .../tests/test_reset_hook.py | 28 +++++++++++++++++++ 4 files changed, 61 insertions(+), 5 deletions(-) create mode 100644 python/paddle/trainer_config_helpers/tests/test_reset_hook.py diff --git a/python/paddle/trainer/config_parser.py b/python/paddle/trainer/config_parser.py index c6c0c9c151..699fe6630a 100644 --- a/python/paddle/trainer/config_parser.py +++ b/python/paddle/trainer/config_parser.py @@ -141,9 +141,9 @@ def init_config_environment( g_add_submodel_suffix=False, # Whether current layer needs to pass the image height and width. - # Default value is true, but if it encounters recurrent_layer_group, - # it will be false. The reason is that image is converted to be sequence, - # image height will be sequence length, and image width will be feature + # Default value is true, but if it encounters recurrent_layer_group, + # it will be false. The reason is that image is converted to be sequence, + # image height will be sequence length, and image width will be feature # length of each timestep. g_pass_height_width=True, ): @@ -1067,7 +1067,7 @@ def cnn_output_size(img_size, filter_size, padding, stride, caffe_mode): return 1 + int(math.ceil(output)) -#calcualte image_size based on output_size for de-convolution (ConvTransLayer). +#calcualte image_size based on output_size for de-convolution (ConvTransLayer). #It is the reverse function of cnn_output_size def cnn_image_size(output_size, filter_size, padding, stride, caffe_mode): img_size = (output_size - 1) * stride + filter_size - 2 * padding @@ -3364,6 +3364,14 @@ def my_fatal(s): logger.critical(s) raise Exception() +_parse_config_hooks = set() +def register_parse_config_hook(f): + """ + Register a hook function for parse_config. parse_config will invoke the hook + at the beginning of parse. This make it possible to reset global state for + for constructing the model. + """ + _parse_config_hooks.add(f) def parse_config(config_file, config_arg_str): ''' @@ -3371,6 +3379,8 @@ def parse_config(config_file, config_arg_str): passed to config script as a dictionary CONFIG_ARGS ''' init_config_environment() + for hook in _parse_config_hooks: + hook() config_args = {} diff --git a/python/paddle/trainer_config_helpers/default_decorators.py b/python/paddle/trainer_config_helpers/default_decorators.py index c01050e338..23a4fa241d 100644 --- a/python/paddle/trainer_config_helpers/default_decorators.py +++ b/python/paddle/trainer_config_helpers/default_decorators.py @@ -78,6 +78,17 @@ class DefaultNameFactory(object): """ pass + def reset(self): + self.__counter__ = 0 + + +_name_factories = [] + +def reset_hook(): + for factory in _name_factories: + factory.reset() + +register_parse_config_hook(reset_hook) def wrap_name_default(name_prefix=None): """ @@ -95,7 +106,9 @@ def wrap_name_default(name_prefix=None): :return: a decorator to set default name :rtype: callable """ - return wrap_param_default(["name"], DefaultNameFactory(name_prefix)) + factory = DefaultNameFactory(name_prefix) + _name_factories.append(factory) + return wrap_param_default(["name"], factory) def wrap_param_attr_default(param_names=None, default_factory=None): diff --git a/python/paddle/trainer_config_helpers/tests/CMakeLists.txt b/python/paddle/trainer_config_helpers/tests/CMakeLists.txt index 6180b2efbc..bff82f7505 100644 --- a/python/paddle/trainer_config_helpers/tests/CMakeLists.txt +++ b/python/paddle/trainer_config_helpers/tests/CMakeLists.txt @@ -4,6 +4,11 @@ add_test(NAME layers_test python ${PROJ_ROOT}/python/paddle/trainer_config_helpers/tests/layers_test.py WORKING_DIRECTORY ${PROJ_ROOT}/python/paddle) +add_test(NAME test_reset_hook + COMMAND ${PROJ_ROOT}/paddle/.set_python_path.sh -d ${PROJ_ROOT}/python/ + python ${PROJ_ROOT}/python/paddle/trainer_config_helpers/tests/test_rest_hook.py + WORKING_DIRECTORY ${PROJ_ROOT}/python/paddle) + if (PROTOBUF_3) add_paddle_exe(protobuf_equal ProtobufEqualMain.cpp) diff --git a/python/paddle/trainer_config_helpers/tests/test_reset_hook.py b/python/paddle/trainer_config_helpers/tests/test_reset_hook.py new file mode 100644 index 0000000000..dc494d0eef --- /dev/null +++ b/python/paddle/trainer_config_helpers/tests/test_reset_hook.py @@ -0,0 +1,28 @@ +# Copyright PaddlePaddle contributors. All Rights Reserved +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import unittest +from paddle.trainer.config_parser import parse_config + +class TestParse(unittest.TestCase): + + def test_parse(self): + a = parse_config( + 'trainer_config_helpers/tests/layers_test_config.py', '') + b = parse_config( + 'trainer_config_helpers/tests/layers_test_config.py', '') + self.assertEqual(a, b) + + +if __name__ == '__main__': + unittest.main() From 6772acab127849f0e94a471d20ef578a339f4179 Mon Sep 17 00:00:00 2001 From: Li Peng Date: Thu, 8 Dec 2016 11:00:02 +0800 Subject: [PATCH 24/60] Update according Yi's comments - Put the 'ARG DEBIAN_FRONTEND=noninteractive' ahead of the first 'apt-get install' invoke. - Add comments to Dockerfiles to explain why we're adding Bazel installation. Signed-off-by: Li Peng --- paddle/scripts/docker/Dockerfile | 7 ++++++- paddle/scripts/docker/Dockerfile.gpu | 7 ++++++- 2 files changed, 12 insertions(+), 2 deletions(-) diff --git a/paddle/scripts/docker/Dockerfile b/paddle/scripts/docker/Dockerfile index a9a72b355e..edb84712d8 100644 --- a/paddle/scripts/docker/Dockerfile +++ b/paddle/scripts/docker/Dockerfile @@ -1,6 +1,7 @@ FROM ubuntu:14.04 MAINTAINER PaddlePaddle Authors +ARG DEBIAN_FRONTEND=noninteractive RUN apt-get update \ && apt-get install -y cmake libprotobuf-dev protobuf-compiler git \ libgoogle-glog-dev libgflags-dev libatlas-dev libatlas3-base g++ m4 python-pip \ @@ -12,7 +13,11 @@ RUN apt-get update \ RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark -ARG DEBIAN_FRONTEND=noninteractive +# cmake tends to hide and blur the dependencies between code modules, as +# noted here https://github.com/PaddlePaddle/Paddle/issues/763. We are +# thinking about using Bazel to fix this problem, e.g., +# https://github.com/PaddlePaddle/Paddle/issues/681#issuecomment-263996102. To +# start the trail of fixing, we add Bazel to our Dockerfiles. RUN apt-get update && apt-get install -y curl software-properties-common \ && add-apt-repository ppa:webupd8team/java \ && echo "oracle-java8-installer shared/accepted-oracle-license-v1-1 select true" | debconf-set-selections \ diff --git a/paddle/scripts/docker/Dockerfile.gpu b/paddle/scripts/docker/Dockerfile.gpu index a147e3840e..5d175e15a7 100644 --- a/paddle/scripts/docker/Dockerfile.gpu +++ b/paddle/scripts/docker/Dockerfile.gpu @@ -1,6 +1,7 @@ FROM nvidia/cuda:7.5-cudnn5-devel-ubuntu14.04 MAINTAINER PaddlePaddle Authors +ARG DEBIAN_FRONTEND=noninteractive RUN apt-get update \ && apt-get install -y cmake libprotobuf-dev protobuf-compiler git \ libgoogle-glog-dev libgflags-dev libatlas-dev libatlas3-base g++ m4 python-pip \ @@ -12,7 +13,11 @@ RUN apt-get update \ RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark -ARG DEBIAN_FRONTEND=noninteractive +# cmake tends to hide and blur the dependencies between code modules, as +# noted here https://github.com/PaddlePaddle/Paddle/issues/763. We are +# thinking about using Bazel to fix this problem, e.g., +# https://github.com/PaddlePaddle/Paddle/issues/681#issuecomment-263996102. To +# start the trail of fixing, we add Bazel to our Dockerfiles. RUN apt-get update && apt-get install -y curl software-properties-common \ && add-apt-repository ppa:webupd8team/java \ && echo "oracle-java8-installer shared/accepted-oracle-license-v1-1 select true" | debconf-set-selections \ From aaecfcc47f0319cb74b80a7e095a8406ce8354ff Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Thu, 8 Dec 2016 11:03:21 +0800 Subject: [PATCH 25/60] Support predicting the samples from sys.stdin --- demo/sentiment/predict.py | 73 ++++++++++++++++++++++----------------- demo/sentiment/predict.sh | 12 +++---- 2 files changed, 47 insertions(+), 38 deletions(-) diff --git a/demo/sentiment/predict.py b/demo/sentiment/predict.py index bc0f6f3126..e01dc6d228 100755 --- a/demo/sentiment/predict.py +++ b/demo/sentiment/predict.py @@ -12,7 +12,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -import os +import os, sys import numpy as np from optparse import OptionParser from py_paddle import swig_paddle, DataProviderConverter @@ -66,35 +66,42 @@ class SentimentPrediction(): for v in open(label_file, 'r'): self.label[int(v.split('\t')[1])] = v.split('\t')[0] - def get_data(self, data_file): + def get_data(self, data): """ Get input data of paddle format. """ - with open(data_file, 'r') as fdata: - for line in fdata: - words = line.strip().split() - word_slot = [ - self.word_dict[w] for w in words if w in self.word_dict - ] - if not word_slot: - print "all words are not in dictionary: %s", line - continue - yield [word_slot] - - def predict(self, data_file): - """ - data_file: file name of input data. - """ - input = self.converter(self.get_data(data_file)) - output = self.network.forwardTest(input) - prob = output[0]["value"] - lab = np.argsort(-prob) - if self.label is None: - print("%s: predicting label is %d" % (data_file, lab[0][0])) - else: - print("%s: predicting label is %s" % - (data_file, self.label[lab[0][0]])) + for line in data: + words = line.strip().split() + word_slot = [ + self.word_dict[w] for w in words if w in self.word_dict + ] + if not word_slot: + print "all words are not in dictionary: %s", line + continue + yield [word_slot] + + def predict(self, batch_size): + + def batch_predict(batch_data): + input = self.converter(self.get_data(batch_data)) + output = self.network.forwardTest(input) + prob = output[0]["value"] + labs = np.argsort(-prob) + for idx, lab in enumerate(labs): + if self.label is None: + print("predicting label is %d" % (lab[0])) + else: + print("predicting label is %s" % + (self.label[lab[0]])) + batch = [] + for line in sys.stdin: + batch.append(line) + if len(batch) == batch_size: + batch_predict(batch) + batch=[] + if len(batch) > 0: + batch_predict(batch) def option_parser(): usage = "python predict.py -n config -w model_dir -d dictionary -i input_file " @@ -119,11 +126,13 @@ def option_parser(): default=None, help="dictionary file") parser.add_option( - "-i", - "--data", + "-c", + "--batch_size", + type="int", action="store", - dest="data", - help="data file to predict") + dest="batch_size", + default=1, + help="the batch size for prediction") parser.add_option( "-w", "--model", @@ -137,13 +146,13 @@ def option_parser(): def main(): options, args = option_parser() train_conf = options.train_conf - data = options.data + batch_size = options.batch_size dict_file = options.dict_file model_path = options.model_path label = options.label swig_paddle.initPaddle("--use_gpu=0") predict = SentimentPrediction(train_conf, dict_file, model_path, label) - predict.predict(data) + predict.predict(batch_size) if __name__ == '__main__': diff --git a/demo/sentiment/predict.sh b/demo/sentiment/predict.sh index 053f23e491..219d2d2025 100755 --- a/demo/sentiment/predict.sh +++ b/demo/sentiment/predict.sh @@ -19,9 +19,9 @@ set -e model=model_output/pass-00002/ config=trainer_config.py label=data/pre-imdb/labels.list -python predict.py \ - -n $config\ - -w $model \ - -b $label \ - -d ./data/pre-imdb/dict.txt \ - -i ./data/aclImdb/test/pos/10007_10.txt +cat ./data/aclImdb/test/pos/10007_10.txt | python predict.py \ + --tconf=$config\ + --model=$model \ + --label=$label \ + --dict=./data/pre-imdb/dict.txt \ + --batch_size=1 From 70e206e8f15eb62c8c0e36c4feb7b585a8bb6e50 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Thu, 8 Dec 2016 10:36:18 +0800 Subject: [PATCH 26/60] follow yanfei's comments --- doc_cn/concepts/use_concepts.rst | 44 ++++++++++++++------------------ doc_cn/faq/index.rst | 2 +- 2 files changed, 20 insertions(+), 26 deletions(-) diff --git a/doc_cn/concepts/use_concepts.rst b/doc_cn/concepts/use_concepts.rst index ea7a5c7d18..5122646d27 100644 --- a/doc_cn/concepts/use_concepts.rst +++ b/doc_cn/concepts/use_concepts.rst @@ -4,9 +4,7 @@ PaddlePaddle 基本使用概念 PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 -单节模式用命令 ``paddle train`` 可以启动一个trainer进程，一个单机训练作业只包括一个trainer进程，单机的所有设备使用，均在单机进程内调度完成。 - -如果数据规模比较大，希望加速训练，可以启动分布式作业。一个分布式作业里包括若干trainer进程和若干Parameter Server（或称pserver）进程。用命令 ``paddle pserver`` 可以启动 pserver 进程，pserver进程用于协调多个trainer进程之间的通信。 +单机模式用命令 ``paddle train`` 可以启动一个trainer进程，单机训练通常只包括一个trainer进程。如果数据规模比较大，希望加速训练，可以启动分布式作业。一个分布式作业里包括若干trainer进程和若干Parameter Server（或称pserver）进程。用命令 ``paddle pserver`` 可以启动 pserver 进程，pserver进程用于协调多个trainer进程之间的通信。 本文首先介绍trainer进程中的一些使用概念，然后介绍pserver进程中概念。 @@ -15,7 +13,7 @@ PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 系统框图 ======== -下图描述了用户使用框图，PaddlePaddle的trainer进程里内嵌了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为DataProvider，通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中, 称为训练配置文件。下面将分别介绍这两部分。 +下图描述了用户使用框图，PaddlePaddle的trainer进程里内嵌了Python解释器，trainer进程可以利用这个解释器执行Python脚本，Python脚本里定义了模型配置、训练算法、以及数据读取函数。其中，数据读取程序往往定义在一个单独Python脚本文件里，被称为数据提供器（DataProvider），通常是一个Python函数。模型配置、训练算法通常定义在另一单独Python文件中, 称为训练配置文件。下面将分别介绍这两部分。 .. graphviz:: @@ -34,8 +32,8 @@ PaddlePaddle是一个深度学习框架，支持单机模式和多机模式。 py -> data_provider [dir="back"]; } -DataProvider -============ +数据提供器 +========== DataProvider是PaddlePaddle系统的数据提供器，将用户的原始数据转换成系统可以识别的数据类型。每当系统需要新的数据训练时, trainer进程会调用DataProvider函数返回数据。当所有数据读取完一轮后，DataProvider返回空数据，通知系统一轮数据读取结束，并且系统每一轮训练开始时会重置DataProvider。需要注意的是，DataProvider是被系统调用，而不是新数据驱动系统，一些随机化噪声添加都应该在DataProvider中完成。 @@ -45,7 +43,7 @@ DataProvider是PaddlePaddle系统的数据提供器，将用户的原始数据 训练配置文件 ============ -训练配置文件主要包括数据传入接口定义(DataConfig)、优化算法(OptimizationConfig)、网络结构(ModelConfig)。 其中数据传入接口定义与DataProvider的关系是：DataProvider里定义数据读取函数，配置文件的DataConfig里指定DataProvider文件名字、生成数据函数接口，请不要混淆。 +训练配置文件主要包括数据源、优化算法、网络结构配置三部分。 其中数据源配置与DataProvider的关系是：DataProvider里定义数据读取函数，训练配置文件的数据源配置中指定DataProvider文件名字、生成数据函数接口，请不要混淆。 一个简单的训练配置文件为： @@ -54,26 +52,22 @@ DataProvider是PaddlePaddle系统的数据提供器，将用户的原始数据 文件开头 ``from paddle.trainer_config_helpers import *`` ，是因为PaddlePaddle配置文件与C++模块通信的最基础协议是protobuf，为了避免用户直接写复杂的protobuf string，我们为用户定以Python接口来配置网络，该Python代码可以生成protobuf包，这就是`trainer_config_helpers`_的作用。因此，在文件的开始，需要import这些函数。 这个包里面包含了模型配置需要的各个模块。 -需要注意的是，这个 ``paddle.trainer_config_helpers`` 包是标准的 Python 包，这意味着用户可以选择自己喜欢的 IDE 或者编辑器来编写配置文件，这个 Python 包注释文档比较完善，并且考虑了 IDE 的代码提示与类型注释。 +下面分别介绍数据源配置、优化算法配置、网络结构配置这三部分该概念。 -下面分别介绍DataConfig、OptimizationConfig、ModelConfig这三部分该概念。 - -DataConfig +数据源配置 ---------- -使用 `PyDataProvider`_ 的函数 ``define_py_data_sources2`` 配置数据源，后缀 2 是Paddle历史遗留问题，因为Paddle之前使用的PyDataProvider性能问题，重构了一个新的 `PyDataProvider`_ 。 - -``define_py_data_sources2`` 里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 +使用 `PyDataProvider`_ 的函数 ``define_py_data_sources2`` 配置数据源。``define_py_data_sources2`` 里通过train_list和test_list指定是训练文件列表和测试文件列表。 如果传入字符串的话，是指一个数据列表文件。这个数据列表文件中包含的是每一个训练或者测试文件的路径。如果传入一个list的话，则会默认生成一个list文件，再传入给train.list或者test.list。 ``module`` 和 ``obj`` 指定了DataProvider的文件名和返回数据的函数名。更详细的使用，请参考 `PyDataProvider`_ 。 -OptimizationConfig ------------------- +优化算法配置 +------------ -通过`settings`_ 接口设置神经网络所使用的训练参数和 `优化算法`_ ，包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 +通过 `settings`_ 接口设置神经网络所使用的训练参数和 `优化算法`_ ，包括学习率、batch_size、优化算法、正则方法等，具体的使用方法请参考 `settings`_ 文档。 -ModelConfig ------------ +网络结构配置 +------------ 神经网络配置主要包括网络连接、激活函数、损失函数、评估器。 @@ -126,11 +120,11 @@ PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trai .. code-block:: bash - paddle pserver --port=5000 --num_gradient_servers=4 --nics='eth0' + paddle pserver --port=5000 --num_gradient_servers=4 --tcp_rdma='tcp' --nics='eth0' -* 指定 pserver 进程端口是 5000 。 -* 有四个训练进程(即 ``--gradient_servers=4`` ，PaddlePaddle同时将 trainer 称作 GradientServer 。因为其为负责提供Gradient) 。 -* 指定以太网类型为TCP网络。 +* ``--port=5000`` : 指定 pserver 进程端口是 5000 。 +* ``--gradient_servers=4`` : 有四个训练进程(PaddlePaddle 将 trainer 也称作 GradientServer ，因为其为负责提供Gradient) 。 +* ``--tcp_rdma='tcp' --nics=`eth0```: 指定以太网类型为TCP网络，指定网络接口名字为eth0。 启动之后 pserver 进程之后，需要启动 trainer 训练进程，在各个机器上运行如下命令\: @@ -140,8 +134,8 @@ PaddlePaddle多机采用经典的 Parameter Server 架构对多个节点的 trai 对于简单的多机协同训练使用上述方式即可。另外，pserver/train 通常在高级情况下，还需要设置下面两个参数\： -* --ports_num\: 一个 pserver 进程共绑定多少个端口用来做稠密更新。默认是1 -* --ports_num_for_sparse\: 一个pserver进程共绑定多少端口用来做稀疏更新，默认是0 +* --ports_num\: 一个 pserver 进程共绑定多少个端口用来做稠密更新，默认是1。 +* --ports_num_for_sparse\: 一个pserver进程共绑定多少端口用来做稀疏更新，默认是0。 使用手工指定端口数量，是因为Paddle的网络通信中，使用了 int32 作为消息长度，比较容易在大模型下溢出。所以，在 pserver 进程中可以启动多个子线程去接受 trainer 的数据，这样单个子线程的长度就不会溢出了。但是这个值不可以调的过大，因为增加这个值，对性能尤其是内存占用有一定的开销，另外稀疏更新的端口如果太大的话，很容易导致某一个参数服务器没有分配到任何参数。 diff --git a/doc_cn/faq/index.rst b/doc_cn/faq/index.rst index 2eba575c2e..2a202527d5 100644 --- a/doc_cn/faq/index.rst +++ b/doc_cn/faq/index.rst @@ -255,7 +255,7 @@ PaddlePaddle的参数使用名字 :code:`name` 作为参数的ID，相同名字 完整源码可参考 `seqToseq `_ 示例。 11. 如何指定GPU设备 ------------------ +------------------- 例如机器上有4块GPU，编号从0开始，指定使用2、3号GPU： From d87b2c11acf00a437925c531890f81c7f38d4eb8 Mon Sep 17 00:00:00 2001 From: xuwei06 Date: Wed, 7 Dec 2016 21:34:56 -0800 Subject: [PATCH 27/60] Fix bug in CMakeList.txt test_rest_hook.py => test_reset_hook.py Change-Id: I84909ade4a1ea2bec4311264626dd000cd6bc86c --- python/paddle/trainer_config_helpers/tests/CMakeLists.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/paddle/trainer_config_helpers/tests/CMakeLists.txt b/python/paddle/trainer_config_helpers/tests/CMakeLists.txt index bff82f7505..d1a9843d32 100644 --- a/python/paddle/trainer_config_helpers/tests/CMakeLists.txt +++ b/python/paddle/trainer_config_helpers/tests/CMakeLists.txt @@ -6,7 +6,7 @@ add_test(NAME layers_test add_test(NAME test_reset_hook COMMAND ${PROJ_ROOT}/paddle/.set_python_path.sh -d ${PROJ_ROOT}/python/ - python ${PROJ_ROOT}/python/paddle/trainer_config_helpers/tests/test_rest_hook.py + python ${PROJ_ROOT}/python/paddle/trainer_config_helpers/tests/test_reset_hook.py WORKING_DIRECTORY ${PROJ_ROOT}/python/paddle) if (PROTOBUF_3) From 20e32ea043ae52262b2eee33f2dee648cddceba7 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Thu, 8 Dec 2016 16:04:56 +0800 Subject: [PATCH 28/60] Follow comments --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 3f5100cbf1..018f93dbf5 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -141,13 +141,13 @@ * 在单层数据的基础上，\ :ref:`glossary_双层RNN`\ 数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。 -* 需要注意的是Paddle目前只支持子序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。例如本里中的两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 +* 需要注意的是PaddlePaddle目前只支持子序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。例如本例中的两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 :ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------ -和示例2中的配置累死，示例3的配置使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ ，实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 +和示例2中的配置类似，示例3的配置使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ ，实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 * 单层\ :ref:`glossary_RNN`\ \: From bc5e0a93bf144f344a5b45aba9c96c968d08f435 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Thu, 8 Dec 2016 16:13:00 +0800 Subject: [PATCH 29/60] Remove ref tags --- doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst | 96 +++++++++---------- 1 file changed, 48 insertions(+), 48 deletions(-) diff --git a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst index 018f93dbf5..9baa0b5780 100644 --- a/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst +++ b/doc_cn/algorithm/rnn/hrnn_rnn_api_compare.rst @@ -4,17 +4,17 @@ 单双层RNN API对比介绍 ##################### -本文以PaddlePaddle的\ :ref:`glossary_双层RNN`\ 单元测试为示例，用多对效果完全相同的、分别使用单双层RNN作为网络配置的模型，来讲解如何使用\ :ref:`glossary_双层RNN`\ 。本文中所有的例子，都只是介绍\ :ref:`glossary_双层RNN`\ 的API接口，并不是使用\ :ref:`glossary_双层RNN`\ 解决实际的问题。如果想要了解\ :ref:`glossary_双层RNN`\ 在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。本文中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 +本文以PaddlePaddle的双层RNN单元测试为示例，用多对效果完全相同的、分别使用单双层RNN作为网络配置的模型，来讲解如何使用双层RNN。本文中所有的例子，都只是介绍双层RNN的API接口，并不是使用双层RNN解决实际的问题。如果想要了解双层RNN在具体问题中的使用，请参考\ :ref:`algo_hrnn_demo`\ 。本文中示例所使用的单元测试文件是\ `test_RecurrentGradientMachine.cpp `_\ 。 示例1：双层RNN，子序列间无Memory ================================ -在\ :ref:`glossary_双层RNN`\ 中的经典情况是将内层的每一个\ :ref:`glossary_sequence`\ 数据，分别进行序列操作；并且内层的序列操作之间独立无依赖，即不需要使用\ :ref:`glossary_Memory`\ 。 +在双层RNN中的经典情况是将内层的每一个时间序列数据，分别进行序列操作；并且内层的序列操作之间独立无依赖，即不需要使用Memory\ 。 -在本示例中，单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是\ :ref:`glossary_RNN`\ 使用两层序列模型，将多句话看成一个整体同时使用encoder压缩。二者语意上完全一致。这组语义相同的示例配置如下： +在本示例中，单层RNN和双层RNN的网络配置，都是将每一句分好词后的句子，使用LSTM作为encoder，压缩成一个向量。区别是RNN使用两层序列模型，将多句话看成一个整体同时使用encoder压缩。二者语意上完全一致。这组语义相同的示例配置如下： -* 单层\ :ref:`glossary_RNN`\: `sequence_layer_group.conf `_ -* :ref:`glossary_双层RNN`\: `sequence_nest_layer_group.conf `_ +* 单层RNN\: `sequence_layer_group.conf `_ +* 双层RNN\: `sequence_nest_layer_group.conf `_ 读取双层序列数据 @@ -22,7 +22,7 @@ 首先，本示例中使用的原始数据如下\: -- 本例中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层\ :ref:`glossary_RNN`\ 网络直接使用。 +- 本例中的原始数据一共有10个样本。每个样本由两部分组成，一个label（此处都为2）和一个已经分词后的句子。这个数据也被单层RNN网络直接使用。 .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg :language: text @@ -33,17 +33,17 @@ .. literalinclude:: ../../../paddle/gserver/tests/Sequence/tour_train_wdseg.nest :language: text -其次，对于两种不同的输入数据类型，不同\ :ref:`glossary_DataProvider`\ 对比如下(`sequenceGen.py `_)\： +其次，对于两种不同的输入数据类型，不同DataProvider对比如下(`sequenceGen.py `_)\： .. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py :language: python :lines: 21-39 :linenos: -- 这是普通的单层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 代码，其说明如下： +- 这是普通的单层时间序列的DataProvider代码，其说明如下： - * :ref:`glossary_DataProvider`\ 共返回两个数据，分别是words和label。即上述代码中的第19行。 - - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层\ :ref:`glossary_sequence`\ 。 + * DataProvider共返回两个数据，分别是words和label。即上述代码中的第19行。 + - words是原始数据中的每一句话，所对应的词表index数组。它是integer_value_sequence类型的，即整数数组。words即为这个数据中的单层时间序列。 - label是原始数据中对于每一句话的分类标签，它是integer_value类型的。 .. literalinclude:: ../../../paddle/gserver/tests/sequenceGen.py @@ -51,17 +51,17 @@ :lines: 42-71 :linenos: -- 对于同样的数据，双层\ :ref:`glossary_sequence`\ 的\ :ref:`glossary_DataProvider`\ 的代码。其说明如下： +- 对于同样的数据，双层时间序列的DataProvider的代码。其说明如下： - - :ref:`glossary_DataProvider`\ 共返回两组数据，分别是sentences和labels。即在双层序列的原始数据中，每一组内的所有句子和labels - - sentences是双层\ :ref:`glossary_sequence`\ 的数据。由于它内部包含了每组数据中的所有句子，且每个句子表示为对应的词表索引数组，因此它是integer_value_sub_sequence 类型的，即双层\ :ref:`glossary_sequence`\ 。 - - labels是每组内每个句子的标签，故而是一个单层\ :ref:`glossary_sequence`\ 。 + - DataProvider共返回两组数据，分别是sentences和labels。即在双层序列的原始数据中，每一组内的所有句子和labels + - sentences是双层时间序列的数据。由于它内部包含了每组数据中的所有句子，且每个句子表示为对应的词表索引数组，因此它是integer_value_sub_sequence 类型的，即双层时间序列。 + - labels是每组内每个句子的标签，故而是一个单层时间序列。 -:ref:`glossary_trainer_config`\ 的模型配置 +模型配置的模型配置 ------------------------------------------ -首先，我们看一下单层\ :ref:`glossary_RNN`\ 的配置。代码中9-15行(高亮部分)即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的\ :ref:`glossary_RNN`\ 处理函数。在这个函数中，\ :ref:`glossary_RNN`\ 对于每一个\ :ref:`glossary_timestep`\ 通过了一个LSTM网络。 +首先，我们看一下单层RNN的配置。代码中9-15行(高亮部分)即为单层RNN序列的使用代码。这里使用了PaddlePaddle预定义好的RNN处理函数。在这个函数中，RNN对于每一个时间步通过了一个LSTM网络。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_layer_group.conf :language: python @@ -70,19 +70,19 @@ :emphasize-lines: 9-15 -其次，我们看一下语义相同的\ :ref:`glossary_双层RNN`\ 的网络配置\: +其次，我们看一下语义相同的双层RNN的网络配置\: -* PaddlePaddle中的许多layer并不在意输入是否是\ :ref:`glossary_sequence`\ ，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个\ :ref:`glossary_timestep`\ 来进行的。 +* PaddlePaddle中的许多layer并不在意输入是否是时间序列，例如\ :code:`embedding_layer`\ 。在这些layer中，所有的操作都是针对每一个时间步来进行的。 -* 在该配置的7-26行(高亮部分)，将双层\ :ref:`glossary_sequence`\ 数据先变换成单层\ :ref:`glossary_sequence`\ 数据，再对每一个单层\ :ref:`glossary_sequence`\ 进行处理。 +* 在该配置的7-26行(高亮部分)，将双层时间序列数据先变换成单层时间序列数据，再对每一个单层时间序列进行处理。 - * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层\ :ref:`glossary_sequence`\ => 单层\ :ref:`glossary_sequence`\ ，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 + * 使用\ :code:`recurrent_group`\ 这个函数进行变换，在变换时需要将输入序列传入。由于我们想要的变换是双层时间序列=> 单层时间序列，所以我们需要将输入数据标记成\ :code:`SubsequenceInput`\ 。 - * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层\ :ref:`glossary_RNN`\ 的配置是等价的。 + * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层RNN的配置是等价的。 -* 与单层\ :ref:`glossary_RNN`\ 的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但和单层\ :ref:`glossary_RNN`\ 不同，我们是对每一个子序列取最后一个元素，因此\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 +* 与单层RNN的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但和单层RNN不同，我们是对每一个子序列取最后一个元素，因此\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。 -* 至此，\ :code:`lstm_last`\ 便和单层\ :ref:`glossary_RNN`\ 配置中的\ :code:`lstm_last`\ 具有相同的结果了。 +* 至此，\ :code:`lstm_last`\ 便和单层RNN配置中的\ :code:`lstm_last`\ 具有相同的结果了。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_layer_group.conf :language: python @@ -90,27 +90,27 @@ :linenos: :emphasize-lines: 7-26 -示例2：:ref:`glossary_双层RNN`，子序列间有\ :ref:`glossary_Memory` -================================================================== +示例2：双层RNN，子序列间有Memory +================================ -本示例意图使用单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 +本示例意图使用单层RNN和双层RNN实现两个完全等价的全连接RNN。 -* 对于单层\ :ref:`glossary_RNN`\ ，输入数据为一个完整的\ :ref:`glossary_sequence`\ ，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。 +* 对于单层RNN，输入数据为一个完整的时间序列，例如\ :code:`[4, 5, 2, 0, 9, 8, 1, 4]`\ 。 -* 对于\ :ref:`glossary_双层RNN`\ ，输入数据为在单层\ :ref:`glossary_RNN`\ 数据里面，任意将一些数据组合成双层\ :ref:`glossary_sequence`\ ，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 +* 对于双层RNN，输入数据为在单层RNN数据里面，任意将一些数据组合成双层时间序列，例如\ :code:`[ [4, 5, 2], [0, 9], [8, 1, 4]]`。 -:ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------- +模型配置的模型配置 +------------------ 我们选取单双层序列配置中的不同部分，来对比分析两者语义相同的原因。 -- 单层\ :ref:`glossary_rnn`\ ：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 +- 单层RNN：过了一个很简单的recurrent_group。每一个时间步，当前的输入y和上一个时间步的输出rnn_state做了一个全链接。 .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn.conf :language: python :lines: 36-48 -- \ :ref:`glossary_双层RNN`\ ，外层memory是一个元素： +- 双层RNN，外层memory是一个元素： - 内层inner_step的recurrent_group和单层序列的几乎一样。除了boot_layer=outer_mem，表示将外层的outer_mem作为内层memory的初始状态。外层outer_step中，outer_mem是一个子句的最后一个向量，即整个双层group是将前一个子句的最后一个向量，作为下一个子句memory的初始状态。 - 从输入数据上看，单双层序列的句子是一样的，只是双层序列将其又做了子序列划分。因此双层序列的配置中，必须将前一个子句的最后一个元素，作为boot_layer传给下一个子句的memory，才能保证和单层序列的配置中“每个时间步都用了上一个时间步的输出结果”一致。 @@ -120,7 +120,7 @@ :lines: 39-66 .. warning:: - PaddlePaddle目前只支持在每个时间步中，Memory的\ :ref:`glossary_sequence`\ 长度一致的情况。 + PaddlePaddle目前只支持在每个时间步中，Memory的时间序列长度一致的情况。 示例3：双层RNN，输入不等长 ========================== @@ -131,32 +131,32 @@ -**输入不等长** 是指recurrent_group的多个输入序列，在每个\ :ref:`glossary_timestep`\ 的子序列长度可以不相等。但序列输出时，需要指定与某一个输入的序列信息是一致的。使用\ :red:`targetInlink`\ 可以指定哪一个输入和输出序列信息一致，默认指定第一个输入。 +**输入不等长** 是指recurrent_group的多个输入序列，在每个时间步的子序列长度可以不相等。但序列输出时，需要指定与某一个输入的序列信息是一致的。使用\ :red:`targetInlink`\ 可以指定哪一个输入和输出序列信息一致，默认指定第一个输入。 示例3的配置分别为\ `单层不等长RNN `_\ 和\ `双层不等长RNN `_\ 。 -示例3对于单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ 数据完全相同。 +示例3对于单层RNN和双层RNN数据完全相同。 -* 对于单层\ :ref:`glossary_RNN`\ 的数据一共有两个样本，他们分别是\ :code:`[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]`\ 和\ :code:`[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]`\ 。对于每一个单层\ :ref:`glossary_RNN`\ 的数据，均有两组特征。 +* 对于单层RNN的数据一共有两个样本，他们分别是\ :code:`[1, 2, 4, 5, 2], [5, 4, 1, 3, 1]`\ 和\ :code:`[0, 2, 2, 5, 0, 1, 2], [1, 5, 4, 2, 3, 6, 1]`\ 。对于每一个单层RNN的数据，均有两组特征。 -* 在单层数据的基础上，\ :ref:`glossary_双层RNN`\ 数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。 +* 在单层数据的基础上，双层RNN数据随意加了一些隔断，例如将第一条数据转化为\ :code:`[[0, 2], [2, 5], [0, 1, 2]],[[1, 5], [4], [2, 3, 6, 1]]`\ 。 -* 需要注意的是PaddlePaddle目前只支持子序列数目一样的多输入\ :ref:`glossary_双层RNN`\ 。例如本例中的两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 +* 需要注意的是PaddlePaddle目前只支持子序列数目一样的多输入双层RNN。例如本例中的两个特征，均有三个子序列。每个子序列长度可以不一致，但是子序列的数目必须一样。 -:ref:`glossary_trainer_config`\ 的模型配置 ------------------------------------------- +模型配置 +-------- -和示例2中的配置类似，示例3的配置使用了单层\ :ref:`glossary_RNN`\ 和\ :ref:`glossary_双层RNN`\ ，实现两个完全等价的全连接\ :ref:`glossary_RNN`\ 。 +和示例2中的配置类似，示例3的配置使用了单层RNN和双层RNN，实现两个完全等价的全连接RNN。 -* 单层\ :ref:`glossary_RNN`\ \: +* 单层RNN\: .. literalinclude:: ../../../paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py :language: python :lines: 42-59 :linenos: -* :ref:`glossary_双层RNN`\ \: +* 双层RNN\ \: .. literalinclude:: ../../../paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py :language: python @@ -179,11 +179,11 @@ TBD Memory ------ -Memory是PaddlePaddle实现 :ref:`glossary_RNN` 时候使用的一个概念。 :ref:`glossary_RNN` 即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。 +Memory是PaddlePaddle实现RNN时候使用的一个概念。RNN即时间递归神经网络，通常要求时间步之间具有一些依赖性，即当前时间步下的神经网络依赖前一个时间步神经网络中某一个神经元输出。如下图所示。 .. graphviz:: glossary_rnn.dot -上图中虚线的连接，即是跨越时间步的网络连接。PaddlePaddle在实现 :ref:`glossary_RNN` 的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的 :ref:`glossary_RNN` 实现便如下图所示。 +上图中虚线的连接，即是跨越时间步的网络连接。PaddlePaddle在实现RNN的时候，将这种跨越时间步的连接用一个特殊的神经网络单元实现。这个神经网络单元就叫Memory。Memory可以缓存上一个时刻某一个神经元的输出，然后在下一个时间步输入给另一个神经元。使用Memory的RNN实现便如下图所示。 .. graphviz:: glossary_rnn_with_memory.dot @@ -194,7 +194,7 @@ Memory是PaddlePaddle实现 :ref:`glossary_RNN` 时候使用的一个概念。 : 时间步 ------ -参考 :ref:`glossary_sequence` 。 +参考时间序列。 .. _glossary_sequence: @@ -217,14 +217,14 @@ RNN RNN 在PaddlePaddle的文档中，一般表示 :code:`Recurrent neural network`，即时间递归神经网络。详细介绍可以参考 `维基百科页面 Recurrent neural network `_ 或者 `中文维基百科页面 `_ 中关于时间递归神经网络的介绍。 -RNN 一般在PaddlePaddle中，指对于一个 :ref:`glossary_sequence` 输入数据，每一个时间步之间的神经网络具有一定的相关性。例如，某一个神经元的一个输入为上一个时间步网络中某一个神经元的输出。或者，从每一个时间步来看，神经网络的网络结构中具有有向环结构。 +RNN 一般在PaddlePaddle中，指对于一个时间序列输入数据，每一个时间步之间的神经网络具有一定的相关性。例如，某一个神经元的一个输入为上一个时间步网络中某一个神经元的输出。或者，从每一个时间步来看，神经网络的网络结构中具有有向环结构。 .. _glossary_双层RNN: 双层RNN ------- -双层RNN顾名思义，即 :ref:`glossary_RNN` 之间有一次嵌套关系。输入数据整体上是一个时间序列，而对于每一个内层特征数据而言，也是一个时间序列。即二维数组，或者数组的数组这个概念。 而双层RNN是可以处理这种输入数据的网络结构。 +双层RNN顾名思义，即RNN之间有一次嵌套关系。输入数据整体上是一个时间序列，而对于每一个内层特征数据而言，也是一个时间序列。即二维数组，或者数组的数组这个概念。 而双层RNN是可以处理这种输入数据的网络结构。 例如，对于段落的文本分类，即将一段话进行分类。我们将一段话看成句子的数组，每个句子又是单词的数组。这便是一种双层RNN的输入数据。而将这个段落的每一句话用lstm编码成一个向量，再对每一句话的编码向量用lstm编码成一个段落的向量。再对这个段落向量进行分类，即为这个双层RNN的网络结构。 From 23d57335f1e76dbc7e63647c092e7ee521844932 Mon Sep 17 00:00:00 2001 From: hanchao Date: Thu, 8 Dec 2016 18:14:25 +0800 Subject: [PATCH 30/60] update cr for pull #734. --- .../tests/configs/run_tests.sh | 10 ++-- .../test_config_parser_for_non_file_config.py | 47 +++++++++++-------- 2 files changed, 35 insertions(+), 22 deletions(-) diff --git a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh index ed2ac6ed18..72dd55bb67 100755 --- a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh +++ b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh @@ -16,14 +16,16 @@ if [ -z $1 ]; then do base_protostr=$protostr/$file new_protostr=$protostr/$file.unittest - diff $base_protostr $new_protostr -u && + diff $base_protostr $new_protostr -u diff $protostr/$file $protostr/$file.non_file_config.unittest -u done else for file in ${configs[*]} do if ! $1 $protostr/$file.protostr $protostr/$file.protostr.unittest; then - diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u && + diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u + fi + if ! $1 $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest; then diff $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest -u fi done @@ -31,7 +33,9 @@ else for file in ${whole_configs[*]} do if ! $1 $protostr/$file.protostr $protostr/$file.protostr.unittest --whole; then - diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u && + diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u + fi + if ! $1 $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest --whole; then diff $protostr/$file.protostr $protostr/$file.protostr.non_file_config.unittest -u fi done diff --git a/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py index 71ee0499d1..87a607acf4 100644 --- a/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py +++ b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py @@ -1,5 +1,5 @@ #!/usr/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,27 +14,36 @@ # limitations under the License. import sys +import re import getopt -whole = False -opts, args = getopt.getopt(sys.argv[1:], "", ["whole"]) -for op, value in opts: - if op == "--whole": - whole = True +def main(print_whole_config, globals, locals): + ''' + this test will all test_config.py + ''' + cmdstr = """from paddle.trainer.config_parser import parse_config\n""" + importstr = "" + functionstr = "" -cmdstr = """ -from paddle.trainer.config_parser import * -from paddle.trainer_config_helpers import * -def configs():\n""" + for line in sys.stdin: + if re.match("^import", line) or re.match("^from.*import", line): + importstr = importstr + line + else: + functionstr = functionstr + " " + line -for line in sys.stdin: - if "import" in line and "from" in line: - continue - cmdstr = cmdstr + " " + line + cmdstr = cmdstr + importstr + """def configs():\n""" + functionstr + #cmdstr = cmdstr + """def configs():\n""" + importstr + functionstr + if print_whole_config: + cmdstr = cmdstr + """print parse_config(configs, "")""" + else: + cmdstr = cmdstr + """print parse_config(configs, "").model_config""" -if whole: - cmdstr = cmdstr + """print parse_config(configs, "")""" -else: - cmdstr = cmdstr + """print parse_config(configs, "").model_config""" + exec(cmdstr, globals, locals) -exec(cmdstr) +if __name__ == '__main__': + whole = False + opts, args = getopt.getopt(sys.argv[1:], "", ["whole"]) + for op, value in opts: + if op == "--whole": + whole = True + main(whole, globals(), locals()) From 957794cd29974acc3d36f201d2754a6e8c096371 Mon Sep 17 00:00:00 2001 From: hanchao Date: Thu, 8 Dec 2016 19:00:10 +0800 Subject: [PATCH 31/60] fix bad ident --- python/paddle/trainer_config_helpers/tests/configs/run_tests.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh index 72dd55bb67..e984ee7062 100755 --- a/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh +++ b/python/paddle/trainer_config_helpers/tests/configs/run_tests.sh @@ -31,7 +31,7 @@ else done for file in ${whole_configs[*]} -do + do if ! $1 $protostr/$file.protostr $protostr/$file.protostr.unittest --whole; then diff $protostr/$file.protostr $protostr/$file.protostr.unittest -u fi From 3adfdf0c583772f5655c31843b802d31b4328c83 Mon Sep 17 00:00:00 2001 From: FoREacH Date: Thu, 8 Dec 2016 13:43:37 +0200 Subject: [PATCH 32/60] Fix nvcc stray character Issue #760 --- CMakeLists.txt | 2 +- paddle/utils/Version.cpp | 6 +++++- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/CMakeLists.txt b/CMakeLists.txt index 0c40e3c6f7..0a44e56719 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -69,7 +69,7 @@ include(coveralls) find_package(Git REQUIRED) # version.cmake will get the current PADDLE_VERSION include(version) -add_definitions(-DPADDLE_VERSION=\"${PADDLE_VERSION}\") +add_definitions(-DPADDLE_VERSION=${PADDLE_VERSION}) if(NOT WITH_GPU) add_definitions(-DPADDLE_ONLY_CPU) diff --git a/paddle/utils/Version.cpp b/paddle/utils/Version.cpp index e706983918..ed4ae6115f 100644 --- a/paddle/utils/Version.cpp +++ b/paddle/utils/Version.cpp @@ -33,7 +33,11 @@ void printVersion(std::ostream& os) { #ifndef PADDLE_VERSION #define PADDLE_VERSION "unknown" #endif - os << "paddle version: " << PADDLE_VERSION << std::endl +// converts macro to string https://gcc.gnu.org/onlinedocs/cpp/Stringification.html +#define xstr(s) str(s) +#define str(s) #s + + os << "paddle version: " << str(PADDLE_VERSION) << std::endl << std::boolalpha << "\t" << "withGpu: " << version::isWithGpu() << std::endl << "\t" From 4690c2e65a91cdf549a344a4b4d4ee284cd6a811 Mon Sep 17 00:00:00 2001 From: livc Date: Thu, 8 Dec 2016 20:47:12 +0800 Subject: [PATCH 33/60] add Chinese version --- .../semantic_role_labeling_cn.md | 201 ++++++++++++++++++ 1 file changed, 201 insertions(+) create mode 100644 doc/tutorials/semantic_role_labeling/semantic_role_labeling_cn.md diff --git a/doc/tutorials/semantic_role_labeling/semantic_role_labeling_cn.md b/doc/tutorials/semantic_role_labeling/semantic_role_labeling_cn.md new file mode 100644 index 0000000000..f3c855a9fd --- /dev/null +++ b/doc/tutorials/semantic_role_labeling/semantic_role_labeling_cn.md @@ -0,0 +1,201 @@ +# 语义角色标注教程 # + +语义角色标注（Semantic role labeling, SRL）是浅语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词参数结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: + + [ A0 他 ] [ AM-MOD 将 ][ AM-NEG 不会 ] [ V 接受] [ A1 任何东西 ] 从 [A2 那些他写的东西中 ]。 + +- V: 动词 +- A0: 接受者 +- A1: 接受的东西 +- A2: 从……接受 +- A3: 属性 +- AM-MOD: 情态动词 +- AM-NEG: 否定 + +给定动词“接受”，句子中的大部分将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 + +到目前为止，大多数成功的SRL系统是建立在某种形式的解析结果之上的，其中在语法结构上使用了预先定义的特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标记问题。 + +## 数据描述 +相关论文[2]采用 CoNLL-2005＆2012 共享任务中设置的数据进行训练和测试。根据数据许可证，演示采用 CoNLL-2005 的测试数据集，可以在网站上找到。 + +用户只需执行以下命令就可以下载并处理原始数据： + +```bash +cd data +./get_data.sh +``` +`data `目录会出现如下几个新的文件： +```bash +conll05st-release：the test data set of CoNll-2005 shared task +test.wsj.words：the Wall Street Journal data sentences +test.wsj.props: the propositional arguments +feature: the extracted features from data set +``` + +## 训练 +### DB-LSTM +请参阅情绪分析的演示以了解有关长期短期记忆单元的更多信息。 + +与在 Sentiment Analysis 演示中使用的 Bidirectional-LSTM 不同，DB-LSTM 采用另一种方法来堆叠LSTM层。首先，标准LSTM以正向处理该序列。该 LSTM 层的输入和输出作为下一个 LSTM 层的输入，并被反向处理。这两个标准 LSTM 层组成一对 LSTM。然后我们堆叠一对对的 LSTM 层后得到深度 LSTM 模型。 + +下图展示了时间扩展的2层 DB-LSTM 网络。 +
+![pic](./network_arch.png) +
+ +### 特征 +两个输入特性在这个管道中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示： +
+![pic](./feature.jpg) +
+ +在这个示例中，相应的标记句子是： + +[ A1 A record date ] has [ AM-NEG n't ] been [ V set ] . + +在演示中, 我们采用上面的特征模板, 包括： `argument`, `predicate`, `ctx-p (p=-1,0,1)`, `mark` 并使用 `B/I/O` 方案来标记每个参数。这些特征和标签存储在 `feature` 文件中, 用`\t`分割。 + +### 数据提供 + +`dataprovider.py` 是一个包装数据的 Python 文件。 函数 `hook()` 定义了网络的数据槽。六个特征和标签都是索引槽。 +``` +def hook(settings, word_dict, label_dict, **kwargs): + settings.word_dict = word_dict + settings.label_dict = label_dict + #all inputs are integral and sequential type + settings.slots = [ + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(predicate_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(2), + integer_value_sequence(len(label_dict))] +``` +相应的数据迭代器如下： +``` +@provider(init_hook=hook, should_shuffle=True, calc_batch_size=get_batch_size, + can_over_batch_size=False, cache=CacheType.CACHE_PASS_IN_MEM) +def process(settings, file_name): + with open(file_name, 'r') as fdata: + for line in fdata: + sentence, predicate, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2, mark, label = \ + line.strip().split('\t') + + words = sentence.split() + sen_len = len(words) + word_slot = [settings.word_dict.get(w, UNK_IDX) for w in words] + + predicate_slot = [settings.predicate_dict.get(predicate)] * sen_len + ctx_n2_slot = [settings.word_dict.get(ctx_n2, UNK_IDX)] * sen_len + ctx_n1_slot = [settings.word_dict.get(ctx_n1, UNK_IDX)] * sen_len + ctx_0_slot = [settings.word_dict.get(ctx_0, UNK_IDX)] * sen_len + ctx_p1_slot = [settings.word_dict.get(ctx_p1, UNK_IDX)] * sen_len + ctx_p2_slot = [settings.word_dict.get(ctx_p2, UNK_IDX)] * sen_len + + marks = mark.split() + mark_slot = [int(w) for w in marks] + + label_list = label.split() + label_slot = [settings.label_dict.get(w) for w in label_list] + yield word_slot, predicate_slot, ctx_n2_slot, ctx_n1_slot, \ + ctx_0_slot, ctx_p1_slot, ctx_p2_slot, mark_slot, label_slot +``` +函数 `process` 产出有8个特征和标签的9个表。 + +### 神经网络配置 + +`db_lstm.py` 是在训练过程中加载字典并定义数据提供程序模块和网络架构的神经网络配置文件。 + +九个 `data_layer` 从数据提供程序加载实例。八个特征分别转换为嵌入，并由`mixed_layer`混合。 深度双向LSTM层提取softmax层的特征。目标函数是标签的交叉熵。 + +### 训练 +训练的脚本是 `train.sh`，用户只需执行: +```bash + ./train.sh +``` +`train.sh` 中的内容： +``` +paddle train \ + --config=./db_lstm.py \ + --use_gpu=0 \ + --log_period=5000 \ + --trainer_count=1 \ + --show_parameter_stats_period=5000 \ + --save_dir=./output \ + --num_passes=10000 \ + --average_test_period=10000000 \ + --init_model_path=./data \ + --load_missing_parameter_strategy=rand \ + --test_all_data_in_one_period=1 \ +2>&1 | tee 'train.log' +``` + +- \--config=./db_lstm.py : 网络配置文件 +- \--use_gpu=false: 使用 CPU 训练（如果已安装 PaddlePaddle GPU版本并想使用 GPU 训练可以设置为true，目前 crf_layer 不支持 GPU） +- \--log_period=500: 每20批(batch)输出日志 +- \--trainer_count=1: 设置线程数（或 GPU 数） +- \--show_parameter_stats_period=5000: 每100批显示参数统计 +- \--save_dir=./output: 模型输出路径 +- \--num_passes=10000: 设置通过数，一次通过意味着PaddlePaddle训练数据集中的所有样本一次 +- \--average_test_period=10000000: 每个 average_test_period 批次对平均参数进行测试 +- \--init_model_path=./data: 参数初始化路径 +- \--load_missing_parameter_strategy=rand: 随机初始不存在的参数 +- \--test_all_data_in_one_period=1: 在一个周期内测试所有数据 + + +训练后，模型将保存在目录`output`中。 我们的训练曲线如下： +
+![pic](./curve.jpg) +
+ +### 测试 +测试脚本是 `test.sh`, 执行: +```bash + ./test.sh +``` +`tesh.sh` 的主要部分： +``` +paddle train \ + --config=./db_lstm.py \ + --model_list=$model_list \ + --job=test \ + --config_args=is_test=1 \ +``` + + - \--config=./db_lstm.py: 网络配置文件 + - \--model_list=$model_list.list: 模型列表文件 + - \--job=test: 指示测试任务 + - \--config_args=is_test=1: 指示测试任务的标记 + - \--test_all_data_in_one_period=1: 在一个周期内测试所有数据 + + +### 预测 +预测脚本是 `predict.sh`，用户只需执行： +```bash + ./predict.sh + +``` +在`predict.sh`中，用户应该提供网络配置文件，模型路径，标签文件，字典文件，特征文件。 +``` +python predict.py + -c $config_file \ + -w $best_model_path \ + -l $label_file \ + -p $predicate_dict_file \ + -d $dict_file \ + -i $input_file \ + -o $output_file +``` + +`predict.py` 是主要的可执行python脚本，其中包括函数：加载模型，加载数据，数据预测。网络模型将输出标签的概率分布。 在演示中，我们使用最大概率的标签作为结果。用户还可以根据概率分布矩阵实现集束搜索或维特比解码。 + +预测后，结果保存在 `predict.res` 中。 + +## 引用 +[1] Martha Palmer, Dan Gildea, and Paul Kingsbury. The Proposition Bank: An Annotated Corpus of Semantic Roles , Computational Linguistics, 31(1), 2005. + +[2] Zhou, Jie, and Wei Xu. "End-to-end learning of semantic role labeling using recurrent neural networks." Proceedings of the Annual Meeting of the Association for Computational Linguistics. 2015. From 38eddca9c26230ab0e6997f2c561cb786c23662b Mon Sep 17 00:00:00 2001 From: FoREacH Date: Thu, 8 Dec 2016 15:12:40 +0200 Subject: [PATCH 34/60] Fixed incorrect macro --- paddle/utils/Version.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/paddle/utils/Version.cpp b/paddle/utils/Version.cpp index ed4ae6115f..84060a1005 100644 --- a/paddle/utils/Version.cpp +++ b/paddle/utils/Version.cpp @@ -37,7 +37,7 @@ void printVersion(std::ostream& os) { #define xstr(s) str(s) #define str(s) #s - os << "paddle version: " << str(PADDLE_VERSION) << std::endl + os << "paddle version: " << xstr(PADDLE_VERSION) << std::endl << std::boolalpha << "\t" << "withGpu: " << version::isWithGpu() << std::endl << "\t" From e9549cbb78b01a570ded56b54cd1f5a0e5f44f12 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Thu, 8 Dec 2016 15:37:00 -0800 Subject: [PATCH 35/60] Change "Baidu, Inc" into "PaddlePaddle Authors" --- LICENSE | 4 ++-- demo/gan/data/download_cifar.sh | 2 +- demo/gan/gan_conf.py | 2 +- demo/gan/gan_conf_image.py | 2 +- demo/gan/gan_trainer.py | 2 +- demo/image_classification/data/download_cifar.sh | 2 +- demo/image_classification/data/process_cifar.py | 2 +- demo/image_classification/image_provider.py | 2 +- demo/image_classification/image_util.py | 2 +- demo/image_classification/predict.sh | 2 +- demo/image_classification/prediction.py | 2 +- demo/image_classification/preprocess.py | 2 +- demo/image_classification/preprocess.sh | 2 +- demo/image_classification/train.sh | 2 +- demo/image_classification/vgg_16_cifar.py | 2 +- demo/introduction/dataprovider.py | 2 +- demo/introduction/evaluate_model.py | 2 +- demo/introduction/train.sh | 2 +- demo/introduction/trainer_config.py | 2 +- demo/mnist/data/generate_list.py | 2 +- demo/mnist/train.sh | 2 +- demo/mnist/vgg_16_mnist.py | 2 +- demo/model_zoo/embedding/extract_para.py | 2 +- demo/model_zoo/embedding/paraconvert.py | 2 +- demo/model_zoo/embedding/pre_DictAndModel.sh | 2 +- demo/model_zoo/resnet/classify.py | 2 +- demo/model_zoo/resnet/example/__init__.py | 2 +- demo/model_zoo/resnet/example/image_list_provider.py | 2 +- demo/model_zoo/resnet/extract_fea_c++.sh | 2 +- demo/model_zoo/resnet/extract_fea_py.sh | 2 +- demo/model_zoo/resnet/get_model.sh | 2 +- demo/model_zoo/resnet/load_feature.py | 2 +- demo/model_zoo/resnet/net_diagram.sh | 2 +- demo/model_zoo/resnet/predict.sh | 2 +- demo/model_zoo/resnet/resnet.py | 2 +- demo/quick_start/api_train.py | 2 +- demo/quick_start/api_train.sh | 2 +- demo/quick_start/data/get_data.sh | 2 +- demo/quick_start/data/proc_from_raw_data/get_data.sh | 2 +- demo/quick_start/data/proc_from_raw_data/preprocess.py | 2 +- demo/quick_start/dataprovider_bow.py | 2 +- demo/quick_start/dataprovider_emb.py | 2 +- demo/quick_start/predict.sh | 2 +- demo/quick_start/train.sh | 2 +- demo/quick_start/trainer_config.bidi-lstm.py | 2 +- demo/quick_start/trainer_config.cnn.py | 2 +- demo/quick_start/trainer_config.db-lstm.py | 2 +- demo/quick_start/trainer_config.emb.py | 2 +- demo/quick_start/trainer_config.lr.py | 2 +- demo/quick_start/trainer_config.lstm.py | 2 +- demo/quick_start/trainer_config.resnet-lstm.py | 2 +- demo/recommendation/common_utils.py | 2 +- demo/recommendation/data/config_generator.py | 2 +- demo/recommendation/data/meta_generator.py | 2 +- demo/recommendation/data/ml_data.sh | 2 +- demo/recommendation/data/split.py | 2 +- demo/recommendation/dataprovider.py | 2 +- demo/recommendation/evaluate.sh | 2 +- demo/recommendation/prediction.py | 2 +- demo/recommendation/preprocess.sh | 2 +- demo/recommendation/run.sh | 2 +- demo/recommendation/trainer_config.py | 2 +- demo/semantic_role_labeling/data/extract_dict_feature.py | 2 +- demo/semantic_role_labeling/data/extract_pairs.py | 2 +- demo/semantic_role_labeling/data/get_data.sh | 2 +- demo/semantic_role_labeling/dataprovider.py | 2 +- demo/semantic_role_labeling/db_lstm.py | 2 +- demo/semantic_role_labeling/predict.py | 2 +- demo/semantic_role_labeling/predict.sh | 2 +- demo/semantic_role_labeling/test.sh | 2 +- demo/semantic_role_labeling/train.sh | 2 +- demo/sentiment/data/get_imdb.sh | 2 +- demo/sentiment/dataprovider.py | 2 +- demo/sentiment/predict.py | 2 +- demo/sentiment/predict.sh | 2 +- demo/sentiment/preprocess.py | 2 +- demo/sentiment/preprocess.sh | 2 +- demo/sentiment/sentiment_net.py | 2 +- demo/sentiment/test.sh | 2 +- demo/sentiment/train.sh | 2 +- demo/sentiment/trainer_config.py | 2 +- demo/seqToseq/data/paraphrase_data.sh | 2 +- demo/seqToseq/data/paraphrase_model.sh | 2 +- demo/seqToseq/data/wmt14_data.sh | 2 +- demo/seqToseq/data/wmt14_model.sh | 2 +- demo/seqToseq/dataprovider.py | 2 +- demo/seqToseq/paraphrase/train.conf | 2 +- demo/seqToseq/paraphrase/train.sh | 2 +- demo/seqToseq/preprocess.py | 2 +- demo/seqToseq/seqToseq_net.py | 2 +- demo/seqToseq/translation/eval_bleu.sh | 2 +- demo/seqToseq/translation/gen.conf | 2 +- demo/seqToseq/translation/gen.sh | 2 +- demo/seqToseq/translation/moses_bleu.sh | 2 +- demo/seqToseq/translation/train.conf | 2 +- demo/seqToseq/translation/train.sh | 2 +- demo/sequence_tagging/data/get_data.sh | 2 +- demo/sequence_tagging/dataprovider.py | 2 +- demo/sequence_tagging/linear_crf.py | 2 +- demo/sequence_tagging/rnn_crf.py | 2 +- doc/api/predict/predict_sample.py | 2 +- doc_cn/cluster/k8s/start_paddle.py | 2 +- paddle/.common_test_util.sh | 2 +- paddle/.set_port.sh | 2 +- paddle/.set_python_path.sh | 2 +- paddle/api/Arguments.cpp | 2 +- paddle/api/ConfigParser.cpp | 2 +- paddle/api/GradientMachine.cpp | 2 +- paddle/api/Internal.h | 2 +- paddle/api/Matrix.cpp | 2 +- paddle/api/PaddleAPI.h | 2 +- paddle/api/PaddleAPIPrivate.h | 2 +- paddle/api/Parameter.cpp | 2 +- paddle/api/ParameterOptimizer.cpp | 2 +- paddle/api/SequenceGenerator.cpp | 2 +- paddle/api/Trainer.cpp | 2 +- paddle/api/Util.cpp | 2 +- paddle/api/Vector.cpp | 2 +- paddle/api/__init__.py | 2 +- paddle/api/paddle_ld_flags.py | 2 +- paddle/api/test/run_tests.sh | 2 +- paddle/api/test/testArguments.py | 2 +- paddle/api/test/testGradientMachine.py | 2 +- paddle/api/test/testMatrix.py | 2 +- paddle/api/test/testTrain.py | 2 +- paddle/api/test/testTrainer.py | 2 +- paddle/api/test/testVector.py | 2 +- paddle/api/test/util.py | 2 +- paddle/cuda/include/hl_activation_functions.h | 2 +- paddle/cuda/include/hl_aggregate.h | 2 +- paddle/cuda/include/hl_avx_functions.h | 2 +- paddle/cuda/include/hl_base.h | 2 +- paddle/cuda/include/hl_batch_transpose.h | 2 +- paddle/cuda/include/hl_cnn.h | 2 +- paddle/cuda/include/hl_cpu_gru.cuh | 2 +- paddle/cuda/include/hl_cpu_lstm.cuh | 2 +- paddle/cuda/include/hl_cpu_matrix_kernel.cuh | 2 +- paddle/cuda/include/hl_cuda.h | 2 +- paddle/cuda/include/hl_cuda.ph | 2 +- paddle/cuda/include/hl_cuda_cublas.h | 2 +- paddle/cuda/include/hl_cuda_cudnn.h | 2 +- paddle/cuda/include/hl_cuda_cudnn.ph | 2 +- paddle/cuda/include/hl_device_functions.cuh | 2 +- paddle/cuda/include/hl_dso_loader.h | 2 +- paddle/cuda/include/hl_functions.h | 2 +- paddle/cuda/include/hl_gpu.h | 2 +- paddle/cuda/include/hl_gpu_functions.cuh | 2 +- paddle/cuda/include/hl_gpu_gru.cuh | 2 +- paddle/cuda/include/hl_gpu_lstm.cuh | 2 +- paddle/cuda/include/hl_gpu_matrix_kernel.cuh | 2 +- paddle/cuda/include/hl_gru_ops.cuh | 2 +- paddle/cuda/include/hl_lstm.h | 2 +- paddle/cuda/include/hl_lstm_ops.cuh | 2 +- paddle/cuda/include/hl_matrix.h | 2 +- paddle/cuda/include/hl_matrix_apply.cuh | 2 +- paddle/cuda/include/hl_matrix_base.cuh | 2 +- paddle/cuda/include/hl_matrix_base_sse.cuh | 2 +- paddle/cuda/include/hl_matrix_ops.cuh | 2 +- paddle/cuda/include/hl_matrix_type.cuh | 2 +- paddle/cuda/include/hl_perturbation_util.cuh | 2 +- paddle/cuda/include/hl_recurrent_apply.cuh | 2 +- paddle/cuda/include/hl_sequence.h | 2 +- paddle/cuda/include/hl_sparse.h | 2 +- paddle/cuda/include/hl_sparse.ph | 2 +- paddle/cuda/include/hl_sse_matrix_kernel.cuh | 2 +- paddle/cuda/include/hl_table_apply.h | 2 +- paddle/cuda/include/hl_tensor_ops.h | 2 +- paddle/cuda/include/hl_thread.ph | 2 +- paddle/cuda/include/hl_time.h | 2 +- paddle/cuda/include/hl_top_k.h | 2 +- paddle/cuda/include/hl_warpctc_wrap.h | 2 +- paddle/cuda/include/stub/hl_aggregate_stub.h | 2 +- paddle/cuda/include/stub/hl_cnn_stub.h | 2 +- paddle/cuda/include/stub/hl_cuda_cublas_stub.h | 2 +- paddle/cuda/include/stub/hl_cuda_cudnn_stub.h | 2 +- paddle/cuda/include/stub/hl_cuda_stub.h | 2 +- paddle/cuda/include/stub/hl_lstm_stub.h | 2 +- paddle/cuda/include/stub/hl_matrix_stub.h | 2 +- paddle/cuda/include/stub/hl_sequence_stub.h | 2 +- paddle/cuda/include/stub/hl_sparse_stub.h | 2 +- paddle/cuda/src/hl_avx_functions.cc | 2 +- paddle/cuda/src/hl_batch_transpose.cu | 2 +- paddle/cuda/src/hl_cpu_functions.cc | 2 +- paddle/cuda/src/hl_cuda_aggregate.cu | 2 +- paddle/cuda/src/hl_cuda_cnn.cu | 2 +- paddle/cuda/src/hl_cuda_cublas.cc | 2 +- paddle/cuda/src/hl_cuda_cudnn.cc | 2 +- paddle/cuda/src/hl_cuda_device.cc | 2 +- paddle/cuda/src/hl_cuda_lstm.cu | 2 +- paddle/cuda/src/hl_cuda_matrix.cu | 2 +- paddle/cuda/src/hl_cuda_sequence.cu | 2 +- paddle/cuda/src/hl_cuda_sparse.cu | 2 +- paddle/cuda/src/hl_cuda_sparse.cuh | 2 +- paddle/cuda/src/hl_cudart_wrap.cc | 2 +- paddle/cuda/src/hl_dso_loader.cc | 2 +- paddle/cuda/src/hl_math.cc | 2 +- paddle/cuda/src/hl_perturbation_util.cu | 2 +- paddle/cuda/src/hl_table_apply.cu | 2 +- paddle/cuda/src/hl_time.cc | 2 +- paddle/cuda/src/hl_top_k.cu | 2 +- paddle/cuda/src/hl_warpctc_wrap.cc | 2 +- paddle/gserver/activations/ActivationFunction.cpp | 2 +- paddle/gserver/activations/ActivationFunction.h | 2 +- paddle/gserver/dataproviders/DataProvider.cpp | 2 +- paddle/gserver/dataproviders/DataProvider.h | 2 +- paddle/gserver/dataproviders/DataProviderGroup.h | 2 +- paddle/gserver/dataproviders/MultiDataProvider.cpp | 2 +- paddle/gserver/dataproviders/MultiDataProvider.h | 2 +- paddle/gserver/dataproviders/ProtoDataProvider.cpp | 2 +- paddle/gserver/dataproviders/ProtoDataProvider.h | 2 +- paddle/gserver/dataproviders/ProtoReader.h | 2 +- paddle/gserver/dataproviders/PyDataProvider.cpp | 2 +- paddle/gserver/dataproviders/PyDataProvider.h | 2 +- paddle/gserver/dataproviders/PyDataProvider2.cpp | 2 +- paddle/gserver/evaluators/CTCErrorEvaluator.cpp | 2 +- paddle/gserver/evaluators/ChunkEvaluator.cpp | 2 +- paddle/gserver/evaluators/Evaluator.cpp | 2 +- paddle/gserver/evaluators/Evaluator.h | 2 +- paddle/gserver/gradientmachines/GradientMachine.cpp | 2 +- paddle/gserver/gradientmachines/GradientMachine.h | 2 +- paddle/gserver/gradientmachines/GradientMachineMode.cpp | 2 +- paddle/gserver/gradientmachines/GradientMachineMode.h | 2 +- paddle/gserver/gradientmachines/MultiGradientMachine.cpp | 2 +- paddle/gserver/gradientmachines/MultiGradientMachine.h | 2 +- paddle/gserver/gradientmachines/MultiNetwork.cpp | 2 +- paddle/gserver/gradientmachines/MultiNetwork.h | 2 +- paddle/gserver/gradientmachines/NeuralNetwork.cpp | 2 +- paddle/gserver/gradientmachines/NeuralNetwork.h | 2 +- paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp | 2 +- paddle/gserver/gradientmachines/ParallelNeuralNetwork.h | 2 +- paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp | 2 +- paddle/gserver/gradientmachines/RecurrentGradientMachine.h | 2 +- paddle/gserver/layers/AddtoLayer.cpp | 2 +- paddle/gserver/layers/AddtoLayer.h | 2 +- paddle/gserver/layers/AgentLayer.cpp | 2 +- paddle/gserver/layers/AgentLayer.h | 2 +- paddle/gserver/layers/AverageLayer.cpp | 2 +- paddle/gserver/layers/AverageLayer.h | 2 +- paddle/gserver/layers/BatchNormBaseLayer.cpp | 2 +- paddle/gserver/layers/BatchNormBaseLayer.h | 2 +- paddle/gserver/layers/BatchNormalizationLayer.cpp | 2 +- paddle/gserver/layers/BatchNormalizationLayer.h | 2 +- paddle/gserver/layers/BilinearInterpLayer.cpp | 2 +- paddle/gserver/layers/BilinearInterpLayer.h | 2 +- paddle/gserver/layers/BlockExpandLayer.cpp | 2 +- paddle/gserver/layers/BlockExpandLayer.h | 2 +- paddle/gserver/layers/CRFDecodingLayer.cpp | 2 +- paddle/gserver/layers/CRFDecodingLayer.h | 2 +- paddle/gserver/layers/CRFLayer.cpp | 2 +- paddle/gserver/layers/CRFLayer.h | 2 +- paddle/gserver/layers/CTCLayer.cpp | 2 +- paddle/gserver/layers/CTCLayer.h | 2 +- paddle/gserver/layers/ConcatenateLayer.cpp | 2 +- paddle/gserver/layers/ContextProjection.cpp | 2 +- paddle/gserver/layers/ContextProjection.h | 2 +- paddle/gserver/layers/ConvBaseLayer.cpp | 2 +- paddle/gserver/layers/ConvBaseLayer.h | 2 +- paddle/gserver/layers/ConvOperator.cpp | 2 +- paddle/gserver/layers/ConvProjection.cpp | 2 +- paddle/gserver/layers/ConvProjection.h | 2 +- paddle/gserver/layers/ConvShiftLayer.cpp | 2 +- paddle/gserver/layers/ConvexCombinationLayer.cpp | 2 +- paddle/gserver/layers/CosSimLayer.cpp | 2 +- paddle/gserver/layers/CosSimLayer.h | 2 +- paddle/gserver/layers/CosSimVecMatLayer.cpp | 2 +- paddle/gserver/layers/CostLayer.cpp | 2 +- paddle/gserver/layers/CostLayer.h | 2 +- paddle/gserver/layers/CudnnBatchNormLayer.cpp | 2 +- paddle/gserver/layers/CudnnBatchNormLayer.h | 2 +- paddle/gserver/layers/CudnnConvLayer.cpp | 2 +- paddle/gserver/layers/CudnnConvLayer.h | 2 +- paddle/gserver/layers/CudnnPoolLayer.cpp | 2 +- paddle/gserver/layers/CudnnPoolLayer.h | 2 +- paddle/gserver/layers/DataLayer.cpp | 2 +- paddle/gserver/layers/DataLayer.h | 2 +- paddle/gserver/layers/DataNormLayer.cpp | 2 +- paddle/gserver/layers/DataNormLayer.h | 2 +- paddle/gserver/layers/DotMulOperator.cpp | 2 +- paddle/gserver/layers/DotMulProjection.cpp | 2 +- paddle/gserver/layers/EosIdCheckLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvBaseLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvBaseLayer.h | 2 +- paddle/gserver/layers/ExpandConvLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvLayer.h | 2 +- paddle/gserver/layers/ExpandConvTransLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvTransLayer.h | 2 +- paddle/gserver/layers/ExpandLayer.cpp | 2 +- paddle/gserver/layers/ExpandLayer.h | 2 +- paddle/gserver/layers/FeatureMapExpandLayer.cpp | 2 +- paddle/gserver/layers/FullMatrixProjection.cpp | 2 +- paddle/gserver/layers/FullMatrixProjection.h | 2 +- paddle/gserver/layers/FullyConnectedLayer.cpp | 2 +- paddle/gserver/layers/FullyConnectedLayer.h | 2 +- paddle/gserver/layers/GatedRecurrentLayer.cpp | 2 +- paddle/gserver/layers/GatedRecurrentLayer.h | 2 +- paddle/gserver/layers/GetOutputLayer.cpp | 2 +- paddle/gserver/layers/GruCompute.cpp | 2 +- paddle/gserver/layers/GruCompute.cu | 2 +- paddle/gserver/layers/GruCompute.h | 2 +- paddle/gserver/layers/GruStepLayer.cpp | 2 +- paddle/gserver/layers/HierarchicalSigmoidLayer.cpp | 2 +- paddle/gserver/layers/HierarchicalSigmoidLayer.h | 2 +- paddle/gserver/layers/IdentityProjection.cpp | 2 +- paddle/gserver/layers/InterpolationLayer.cpp | 2 +- paddle/gserver/layers/Layer.cpp | 2 +- paddle/gserver/layers/Layer.h | 2 +- paddle/gserver/layers/LinearChainCRF.cpp | 2 +- paddle/gserver/layers/LinearChainCRF.h | 2 +- paddle/gserver/layers/LinearChainCTC.cpp | 2 +- paddle/gserver/layers/LinearChainCTC.h | 2 +- paddle/gserver/layers/LstmCompute.cpp | 2 +- paddle/gserver/layers/LstmCompute.cu | 2 +- paddle/gserver/layers/LstmCompute.h | 2 +- paddle/gserver/layers/LstmLayer.cpp | 2 +- paddle/gserver/layers/LstmLayer.h | 2 +- paddle/gserver/layers/LstmStepLayer.cpp | 2 +- paddle/gserver/layers/MDLstmLayer.cpp | 2 +- paddle/gserver/layers/MaxIdLayer.cpp | 2 +- paddle/gserver/layers/MaxLayer.cpp | 2 +- paddle/gserver/layers/MaxLayer.h | 2 +- paddle/gserver/layers/MaxOutLayer.cpp | 2 +- paddle/gserver/layers/MaxOutLayer.h | 2 +- paddle/gserver/layers/MixedLayer.cpp | 2 +- paddle/gserver/layers/MixedLayer.h | 2 +- paddle/gserver/layers/MultinomialSampler.cpp | 2 +- paddle/gserver/layers/MultinomialSampler.h | 2 +- paddle/gserver/layers/MultiplexLayer.cpp | 2 +- paddle/gserver/layers/NCELayer.cpp | 2 +- paddle/gserver/layers/NormLayer.cpp | 2 +- paddle/gserver/layers/NormLayer.h | 2 +- paddle/gserver/layers/NormProjectionLayer.cpp | 2 +- paddle/gserver/layers/NormProjectionLayer.h | 2 +- paddle/gserver/layers/Operator.cpp | 2 +- paddle/gserver/layers/Operator.h | 2 +- paddle/gserver/layers/OuterProdLayer.cpp | 2 +- paddle/gserver/layers/ParameterReluLayer.cpp | 2 +- paddle/gserver/layers/ParameterReluLayer.h | 2 +- paddle/gserver/layers/PoolLayer.cpp | 2 +- paddle/gserver/layers/PoolLayer.h | 2 +- paddle/gserver/layers/PoolProjection.cpp | 2 +- paddle/gserver/layers/PoolProjection.h | 2 +- paddle/gserver/layers/PoolProjectionLayer.cpp | 2 +- paddle/gserver/layers/PoolProjectionLayer.h | 2 +- paddle/gserver/layers/PowerLayer.cpp | 2 +- paddle/gserver/layers/PrintLayer.cpp | 2 +- paddle/gserver/layers/Projection.cpp | 2 +- paddle/gserver/layers/Projection.h | 2 +- paddle/gserver/layers/RecurrentLayer.cpp | 2 +- paddle/gserver/layers/RecurrentLayerGroup.cpp | 2 +- paddle/gserver/layers/ResizeLayer.cpp | 2 +- paddle/gserver/layers/SamplingIdLayer.cpp | 2 +- paddle/gserver/layers/ScalingLayer.cpp | 2 +- paddle/gserver/layers/ScalingProjection.cpp | 2 +- paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp | 2 +- paddle/gserver/layers/SelectiveFullyConnectedLayer.h | 2 +- paddle/gserver/layers/SequenceConcatLayer.cpp | 2 +- paddle/gserver/layers/SequenceLastInstanceLayer.cpp | 2 +- paddle/gserver/layers/SequencePoolLayer.cpp | 2 +- paddle/gserver/layers/SequencePoolLayer.h | 2 +- paddle/gserver/layers/SequenceReshapeLayer.cpp | 2 +- paddle/gserver/layers/SequenceToBatch.cpp | 2 +- paddle/gserver/layers/SequenceToBatch.h | 2 +- paddle/gserver/layers/SlopeInterceptLayer.cpp | 2 +- paddle/gserver/layers/SpatialPyramidPoolLayer.cpp | 2 +- paddle/gserver/layers/SpatialPyramidPoolLayer.h | 2 +- paddle/gserver/layers/SubSequenceLayer.cpp | 2 +- paddle/gserver/layers/SumToOneNormLayer.cpp | 2 +- paddle/gserver/layers/TableProjection.cpp | 2 +- paddle/gserver/layers/TableProjection.h | 2 +- paddle/gserver/layers/TensorLayer.cpp | 2 +- paddle/gserver/layers/TensorLayer.h | 2 +- paddle/gserver/layers/TransLayer.cpp | 2 +- paddle/gserver/layers/TransLayer.h | 2 +- paddle/gserver/layers/TransposedFullMatrixProjection.cpp | 2 +- paddle/gserver/layers/ValidationLayer.cpp | 2 +- paddle/gserver/layers/ValidationLayer.h | 2 +- paddle/gserver/layers/WarpCTCLayer.cpp | 2 +- paddle/gserver/layers/WarpCTCLayer.h | 2 +- paddle/gserver/tests/LayerGradUtil.cpp | 2 +- paddle/gserver/tests/LayerGradUtil.h | 2 +- paddle/gserver/tests/TestUtil.cpp | 2 +- paddle/gserver/tests/TestUtil.h | 2 +- paddle/gserver/tests/__init__.py | 2 +- paddle/gserver/tests/concat_dotmul_a.conf | 2 +- paddle/gserver/tests/concat_dotmul_b.conf | 2 +- paddle/gserver/tests/concat_fullmatrix_a.conf | 2 +- paddle/gserver/tests/concat_fullmatrix_b.conf | 2 +- paddle/gserver/tests/concat_table_a.conf | 2 +- paddle/gserver/tests/concat_table_b.conf | 2 +- paddle/gserver/tests/img_conv_a.conf | 2 +- paddle/gserver/tests/img_conv_b.conf | 2 +- paddle/gserver/tests/img_conv_c.conf | 2 +- paddle/gserver/tests/img_pool_a.conf | 2 +- paddle/gserver/tests/img_pool_b.conf | 2 +- paddle/gserver/tests/pyDataProvider.py | 2 +- paddle/gserver/tests/pyDataProvider/trainer.conf | 2 +- paddle/gserver/tests/rnn_data_provider.py | 2 +- paddle/gserver/tests/sequenceGen.py | 2 +- paddle/gserver/tests/sequence_layer_group.conf | 2 +- paddle/gserver/tests/sequence_nest_layer_group.conf | 2 +- paddle/gserver/tests/sequence_nest_rnn.conf | 2 +- paddle/gserver/tests/sequence_nest_rnn_multi_input.conf | 2 +- .../tests/sequence_nest_rnn_multi_unequalength_inputs.py | 2 +- paddle/gserver/tests/sequence_rnn.conf | 2 +- paddle/gserver/tests/sequence_rnn_multi_input.conf | 2 +- .../gserver/tests/sequence_rnn_multi_unequalength_inputs.py | 2 +- paddle/gserver/tests/test_ActivationGrad.cpp | 2 +- paddle/gserver/tests/test_BatchNorm.cpp | 2 +- paddle/gserver/tests/test_ConvTrans.cpp | 2 +- paddle/gserver/tests/test_ConvUnify.cpp | 2 +- paddle/gserver/tests/test_Evaluator.cpp | 2 +- paddle/gserver/tests/test_LayerGrad.cpp | 2 +- paddle/gserver/tests/test_LinearChainCRF.cpp | 2 +- paddle/gserver/tests/test_MultinomialSampler.cpp | 2 +- paddle/gserver/tests/test_NetworkCompare.cpp | 2 +- paddle/gserver/tests/test_ProtoDataProvider.cpp | 2 +- paddle/gserver/tests/test_PyDataProvider.cpp | 2 +- paddle/gserver/tests/test_PyDataProvider2.cpp | 2 +- paddle/gserver/tests/test_PyDataProvider2.py | 2 +- paddle/gserver/tests/test_RecurrentGradientMachine.cpp | 2 +- paddle/gserver/tests/test_RecurrentLayer.cpp | 2 +- paddle/gserver/tests/test_SelectiveFCLayer.cpp | 2 +- paddle/gserver/tests/test_WarpCTCLayer.cpp | 2 +- paddle/math/Allocator.h | 2 +- paddle/math/BaseMatrix.cu | 2 +- paddle/math/BaseMatrix.h | 2 +- paddle/math/CpuSparseMatrix.cpp | 2 +- paddle/math/CpuSparseMatrix.h | 2 +- paddle/math/ExecViaCpu.h | 2 +- paddle/math/MathFunctions.cpp | 2 +- paddle/math/MathFunctions.h | 2 +- paddle/math/MathUtils.cpp | 2 +- paddle/math/MathUtils.h | 2 +- paddle/math/Matrix.cpp | 2 +- paddle/math/Matrix.h | 2 +- paddle/math/MatrixBitCode.cpp | 2 +- paddle/math/MemoryHandle.cpp | 2 +- paddle/math/MemoryHandle.h | 2 +- paddle/math/PoolAllocator.cpp | 2 +- paddle/math/PoolAllocator.h | 2 +- paddle/math/SIMDFunctions.cpp | 2 +- paddle/math/SIMDFunctions.h | 2 +- paddle/math/SparseMatrix.cpp | 2 +- paddle/math/SparseMatrix.h | 2 +- paddle/math/SparseRowMatrix.cpp | 2 +- paddle/math/SparseRowMatrix.h | 2 +- paddle/math/Storage.cpp | 2 +- paddle/math/Storage.h | 2 +- paddle/math/TensorApply.h | 2 +- paddle/math/TensorAssign.h | 2 +- paddle/math/TensorEvaluate.h | 2 +- paddle/math/TensorExpression.h | 2 +- paddle/math/TrainingAlgorithmOp.cu | 2 +- paddle/math/TrainingAlgorithmOp.h | 2 +- paddle/math/Vector.cpp | 2 +- paddle/math/Vector.h | 2 +- paddle/math/tests/OriginalOptimizerApi.h | 2 +- paddle/math/tests/PerfUtils.h | 2 +- paddle/math/tests/TensorCheck.h | 2 +- paddle/math/tests/TestUtils.h | 2 +- paddle/math/tests/test_Allocator.cpp | 2 +- paddle/math/tests/test_BaseMatrix.cpp | 2 +- paddle/math/tests/test_CpuGpuVector.cpp | 2 +- paddle/math/tests/test_ExecViaCpu.cpp | 2 +- paddle/math/tests/test_FPException.cpp | 2 +- paddle/math/tests/test_GpuProfiler.cpp | 2 +- paddle/math/tests/test_Matrix.cpp | 2 +- paddle/math/tests/test_SIMDFunctions.cpp | 2 +- paddle/math/tests/test_SparseMatrix.cpp | 2 +- paddle/math/tests/test_Tensor.cu | 2 +- paddle/math/tests/test_TrainingAlgorithm.cpp | 2 +- paddle/math/tests/test_batchTranspose.cpp | 2 +- paddle/math/tests/test_lazyAssign.cu | 2 +- paddle/math/tests/test_matrixCompare.cpp | 2 +- paddle/math/tests/test_matrixUtil.h | 2 +- paddle/math/tests/test_perturbation.cpp | 2 +- paddle/math/tests/test_sparseMatrixCompare.cpp | 2 +- paddle/parameter/Argument.cpp | 2 +- paddle/parameter/Argument.h | 2 +- paddle/parameter/AverageOptimizer.cpp | 2 +- paddle/parameter/AverageOptimizer.h | 2 +- paddle/parameter/FirstOrderOptimizer.cpp | 2 +- paddle/parameter/FirstOrderOptimizer.h | 2 +- paddle/parameter/LearningRateScheduler.cpp | 2 +- paddle/parameter/LearningRateScheduler.h | 2 +- paddle/parameter/OptimizerFunctions.cpp | 2 +- paddle/parameter/OptimizerFunctions.h | 2 +- paddle/parameter/OptimizerWithRegularizer.cpp | 2 +- paddle/parameter/OptimizerWithRegularizer.h | 2 +- paddle/parameter/ParallelParameter.cpp | 2 +- paddle/parameter/ParallelParameter.h | 2 +- paddle/parameter/Parameter.cpp | 2 +- paddle/parameter/Parameter.h | 2 +- paddle/parameter/ParameterOptimizer.cpp | 2 +- paddle/parameter/ParameterOptimizer.h | 2 +- paddle/parameter/ParameterUpdateFunctions.cpp | 2 +- paddle/parameter/ParameterUpdateFunctions.h | 2 +- paddle/parameter/ParameterUpdaterBase.cpp | 2 +- paddle/parameter/ParameterUpdaterBase.h | 2 +- paddle/parameter/ParameterUpdaterHook.cpp | 2 +- paddle/parameter/ParameterUpdaterHook.h | 2 +- paddle/parameter/Regularizer.cpp | 2 +- paddle/parameter/Regularizer.h | 2 +- paddle/parameter/Weight.cpp | 2 +- paddle/parameter/Weight.h | 2 +- paddle/parameter/tests/test_common.cpp | 2 +- paddle/pserver/BaseClient.cpp | 2 +- paddle/pserver/BaseClient.h | 2 +- paddle/pserver/LightNetwork.cpp | 2 +- paddle/pserver/LightNetwork.h | 2 +- paddle/pserver/ParameterClient2.cpp | 2 +- paddle/pserver/ParameterClient2.h | 2 +- paddle/pserver/ParameterServer2.cpp | 2 +- paddle/pserver/ParameterServer2.h | 2 +- paddle/pserver/ParameterServer2Main.cpp | 2 +- paddle/pserver/ProtoServer.cpp | 2 +- paddle/pserver/ProtoServer.h | 2 +- paddle/pserver/RDMANetwork.h | 2 +- paddle/pserver/SocketChannel.cpp | 2 +- paddle/pserver/SocketChannel.h | 2 +- paddle/pserver/SparseParameterDistribution.cpp | 2 +- paddle/pserver/SparseParameterDistribution.h | 2 +- paddle/pserver/test/SocketTest.cpp | 2 +- paddle/pserver/test/test_ParameterServer2.cpp | 2 +- paddle/pserver/test/test_ProtoServer.cpp | 2 +- paddle/pserver/test/test_ProtoServer.sh | 2 +- paddle/py_paddle/__init__.py | 2 +- paddle/py_paddle/dataprovider_converter.py | 2 +- paddle/py_paddle/util.py | 2 +- paddle/scripts/cluster_train/conf.py | 2 +- paddle/scripts/cluster_train/paddle.py | 2 +- paddle/setup.py.in | 2 +- paddle/trainer/MergeModel.cpp | 2 +- paddle/trainer/ParamUtil.cpp | 2 +- paddle/trainer/ParamUtil.h | 2 +- paddle/trainer/ParameterUpdater.cpp | 2 +- paddle/trainer/ParameterUpdater.h | 2 +- paddle/trainer/RemoteParameterUpdater.cpp | 2 +- paddle/trainer/RemoteParameterUpdater.h | 2 +- paddle/trainer/Tester.cpp | 2 +- paddle/trainer/Tester.h | 2 +- paddle/trainer/TesterConfig.h | 2 +- paddle/trainer/ThreadParameterUpdater.cpp | 2 +- paddle/trainer/ThreadParameterUpdater.h | 2 +- paddle/trainer/Trainer.cpp | 2 +- paddle/trainer/Trainer.h | 2 +- paddle/trainer/TrainerBenchmark.cpp | 2 +- paddle/trainer/TrainerConfigHelper.cpp | 2 +- paddle/trainer/TrainerConfigHelper.h | 2 +- paddle/trainer/TrainerInternal.cpp | 2 +- paddle/trainer/TrainerInternal.h | 2 +- paddle/trainer/TrainerInternalConfig.cpp | 2 +- paddle/trainer/TrainerInternalConfig.h | 2 +- paddle/trainer/TrainerMain.cpp | 2 +- paddle/trainer/tests/__init__.py | 2 +- paddle/trainer/tests/chunking.conf | 2 +- paddle/trainer/tests/config_parser_test.py | 2 +- paddle/trainer/tests/gen_proto_data.py | 2 +- paddle/trainer/tests/sample_trainer_config.conf | 2 +- paddle/trainer/tests/sample_trainer_config_hsigmoid.conf | 2 +- paddle/trainer/tests/sample_trainer_config_opt_a.conf | 2 +- paddle/trainer/tests/sample_trainer_config_opt_b.conf | 2 +- paddle/trainer/tests/sample_trainer_config_parallel.conf | 2 +- paddle/trainer/tests/sample_trainer_config_qb_rnn.conf | 2 +- paddle/trainer/tests/sample_trainer_config_rnn.conf | 2 +- paddle/trainer/tests/sample_trainer_nest_rnn_gen.conf | 2 +- paddle/trainer/tests/sample_trainer_rnn_gen.conf | 2 +- paddle/trainer/tests/testPyDataWrapper.py | 2 +- paddle/trainer/tests/test_Compare.cpp | 2 +- paddle/trainer/tests/test_CompareSparse.cpp | 2 +- paddle/trainer/tests/test_CompareTwoNets.cpp | 2 +- paddle/trainer/tests/test_CompareTwoOpts.cpp | 2 +- paddle/trainer/tests/test_Prediction.cpp | 2 +- paddle/trainer/tests/test_PyDataProviderWrapper.cpp | 2 +- paddle/trainer/tests/test_Trainer.cpp | 2 +- paddle/trainer/tests/test_TrainerOnePass.cpp | 2 +- paddle/trainer/tests/test_config.conf | 2 +- paddle/trainer/tests/test_recurrent_machine_generation.cpp | 2 +- paddle/utils/BarrierStat.cpp | 2 +- paddle/utils/BarrierStat.h | 2 +- paddle/utils/ClassRegistrar.h | 2 +- paddle/utils/CommandLineParser.cpp | 2 +- paddle/utils/CommandLineParser.h | 2 +- paddle/utils/CompilerMacros.h | 2 +- paddle/utils/CustomStackTrace.cpp | 2 +- paddle/utils/CustomStackTrace.h | 2 +- paddle/utils/DisableCopy.h | 2 +- paddle/utils/Excepts.cpp | 2 +- paddle/utils/Excepts.h | 2 +- paddle/utils/Flags.cpp | 2 +- paddle/utils/Flags.h | 2 +- paddle/utils/GlobalConstants.cpp | 2 +- paddle/utils/GlobalConstants.h | 2 +- paddle/utils/Locks.h | 2 +- paddle/utils/Logging.cpp | 2 +- paddle/utils/Logging.h | 2 +- paddle/utils/PythonUtil.cpp | 2 +- paddle/utils/PythonUtil.h | 2 +- paddle/utils/Queue.h | 2 +- paddle/utils/Stat.cpp | 2 +- paddle/utils/Stat.h | 2 +- paddle/utils/StringUtil.cpp | 2 +- paddle/utils/StringUtil.h | 2 +- paddle/utils/Thread.h | 2 +- paddle/utils/ThreadLocal.cpp | 2 +- paddle/utils/ThreadLocal.h | 2 +- paddle/utils/TypeDefs.h | 2 +- paddle/utils/Util.cpp | 2 +- paddle/utils/Util.h | 2 +- paddle/utils/Version.cpp | 2 +- paddle/utils/Version.h | 2 +- paddle/utils/arch/linux/Locks.cpp | 2 +- paddle/utils/arch/osx/Locks.cpp | 2 +- paddle/utils/tests/test_CommandLineParser.cpp | 2 +- paddle/utils/tests/test_CustomStackTrace.cpp | 2 +- paddle/utils/tests/test_CustomStackTracePrint.cpp | 2 +- paddle/utils/tests/test_Logging.cpp | 2 +- paddle/utils/tests/test_SpinLock.cpp | 2 +- paddle/utils/tests/test_StringUtils.cpp | 2 +- paddle/utils/tests/test_Thread.cpp | 2 +- paddle/utils/tests/test_ThreadBarrier.cpp | 2 +- proto/DataConfig.proto.m4 | 2 +- proto/DataFormat.proto.m4 | 2 +- proto/ModelConfig.proto.m4 | 2 +- proto/ParameterConfig.proto.m4 | 2 +- proto/ParameterService.proto.m4 | 2 +- proto/TrainerConfig.proto.m4 | 2 +- python/paddle/__init__.py | 2 +- python/paddle/proto/__init__.py | 2 +- python/paddle/trainer/PyDataProvider2.py | 2 +- python/paddle/trainer/PyDataProviderWrapper.py | 2 +- python/paddle/trainer/__init__.py | 2 +- python/paddle/trainer/config_parser.py | 2 +- python/paddle/trainer/config_parser_extension.py | 2 +- python/paddle/trainer/recurrent_units.py | 2 +- python/paddle/trainer_config_helpers/__init__.py | 2 +- python/paddle/trainer_config_helpers/activations.py | 2 +- python/paddle/trainer_config_helpers/attrs.py | 2 +- python/paddle/trainer_config_helpers/data_sources.py | 2 +- python/paddle/trainer_config_helpers/default_decorators.py | 2 +- python/paddle/trainer_config_helpers/evaluators.py | 2 +- python/paddle/trainer_config_helpers/layers.py | 2 +- python/paddle/trainer_config_helpers/math.py | 2 +- python/paddle/trainer_config_helpers/networks.py | 2 +- python/paddle/trainer_config_helpers/optimizers.py | 2 +- python/paddle/trainer_config_helpers/poolings.py | 2 +- .../paddle/trainer_config_helpers/tests/ProtobufEqualMain.cpp | 2 +- python/paddle/trainer_config_helpers/tests/layers_test.py | 2 +- .../paddle/trainer_config_helpers/tests/layers_test_config.py | 2 +- python/paddle/trainer_config_helpers/utils.py | 2 +- python/paddle/utils/__init__.py | 2 +- python/paddle/utils/dump_config.py | 2 +- python/paddle/utils/image_util.py | 2 +- python/paddle/utils/make_model_diagram.py | 2 +- python/paddle/utils/plotcurve.py | 2 +- python/paddle/utils/predefined_net.py | 2 +- python/paddle/utils/preprocess_img.py | 2 +- python/paddle/utils/preprocess_util.py | 2 +- python/paddle/utils/show_pb.py | 2 +- python/paddle/utils/torch2paddle.py | 2 +- 660 files changed, 661 insertions(+), 661 deletions(-) diff --git a/LICENSE b/LICENSE index 2ff3140db0..e77bd090ee 100644 --- a/LICENSE +++ b/LICENSE @@ -1,4 +1,4 @@ -Copyright (c) 2016 Baidu, Inc. All Rights Reserved +Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved Apache License Version 2.0, January 2004 @@ -188,7 +188,7 @@ Copyright (c) 2016 Baidu, Inc. All Rights Reserved same "printed page" as the copyright notice for easier identification within third-party archives. - Copyright (c) 2016 Baidu, Inc. All Rights Reserve. + Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/demo/gan/data/download_cifar.sh b/demo/gan/data/download_cifar.sh index ea3be594cd..32e73b3d8e 100755 --- a/demo/gan/data/download_cifar.sh +++ b/demo/gan/data/download_cifar.sh @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/gan/gan_conf.py b/demo/gan/gan_conf.py index 05eee3a9b9..58ba9dde58 100644 --- a/demo/gan/gan_conf.py +++ b/demo/gan/gan_conf.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/gan/gan_conf_image.py b/demo/gan/gan_conf_image.py index dc5910e9f0..5c2b140537 100644 --- a/demo/gan/gan_conf_image.py +++ b/demo/gan/gan_conf_image.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/gan/gan_trainer.py b/demo/gan/gan_trainer.py index 72699952b9..a8c1bd0414 100644 --- a/demo/gan/gan_trainer.py +++ b/demo/gan/gan_trainer.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/data/download_cifar.sh b/demo/image_classification/data/download_cifar.sh index ca9b0b5c90..52e82d0d98 100755 --- a/demo/image_classification/data/download_cifar.sh +++ b/demo/image_classification/data/download_cifar.sh @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/data/process_cifar.py b/demo/image_classification/data/process_cifar.py index b235010e4e..db6666189e 100644 --- a/demo/image_classification/data/process_cifar.py +++ b/demo/image_classification/data/process_cifar.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/image_provider.py b/demo/image_classification/image_provider.py index 28bf1bb02c..87eed5eebd 100644 --- a/demo/image_classification/image_provider.py +++ b/demo/image_classification/image_provider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/image_util.py b/demo/image_classification/image_util.py index b5c6431c06..f09605394a 100644 --- a/demo/image_classification/image_util.py +++ b/demo/image_classification/image_util.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/predict.sh b/demo/image_classification/predict.sh index 35ffae6c8c..9d5785c9a1 100755 --- a/demo/image_classification/predict.sh +++ b/demo/image_classification/predict.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/prediction.py b/demo/image_classification/prediction.py index 6a47bd5851..9a86aafcb2 100755 --- a/demo/image_classification/prediction.py +++ b/demo/image_classification/prediction.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/preprocess.py b/demo/image_classification/preprocess.py index 10b9c1691b..2947ad239c 100755 --- a/demo/image_classification/preprocess.py +++ b/demo/image_classification/preprocess.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/preprocess.sh b/demo/image_classification/preprocess.sh index e3e86ff106..c7396c6393 100755 --- a/demo/image_classification/preprocess.sh +++ b/demo/image_classification/preprocess.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/train.sh b/demo/image_classification/train.sh index db0a057bf3..6fc11caf1c 100755 --- a/demo/image_classification/train.sh +++ b/demo/image_classification/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/image_classification/vgg_16_cifar.py b/demo/image_classification/vgg_16_cifar.py index 58ceff5fc2..8ee4a64c15 100755 --- a/demo/image_classification/vgg_16_cifar.py +++ b/demo/image_classification/vgg_16_cifar.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/introduction/dataprovider.py b/demo/introduction/dataprovider.py index 8515022e18..03c920cc34 100644 --- a/demo/introduction/dataprovider.py +++ b/demo/introduction/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/introduction/evaluate_model.py b/demo/introduction/evaluate_model.py index ca4a187273..eeda43c5c8 100755 --- a/demo/introduction/evaluate_model.py +++ b/demo/introduction/evaluate_model.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: UTF-8 -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/introduction/train.sh b/demo/introduction/train.sh index 06db8edd10..b7bbb90ddd 100755 --- a/demo/introduction/train.sh +++ b/demo/introduction/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/introduction/trainer_config.py b/demo/introduction/trainer_config.py index 7c838c1a8f..41cebcf6e1 100644 --- a/demo/introduction/trainer_config.py +++ b/demo/introduction/trainer_config.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/mnist/data/generate_list.py b/demo/mnist/data/generate_list.py index d880721f94..49981cc7a9 100644 --- a/demo/mnist/data/generate_list.py +++ b/demo/mnist/data/generate_list.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/mnist/train.sh b/demo/mnist/train.sh index 084b32ac39..da90cd749a 100755 --- a/demo/mnist/train.sh +++ b/demo/mnist/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/mnist/vgg_16_mnist.py b/demo/mnist/vgg_16_mnist.py index f9e89bc588..a819b391c6 100644 --- a/demo/mnist/vgg_16_mnist.py +++ b/demo/mnist/vgg_16_mnist.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/embedding/extract_para.py b/demo/model_zoo/embedding/extract_para.py index 47e06fae9c..570b90c1f7 100755 --- a/demo/model_zoo/embedding/extract_para.py +++ b/demo/model_zoo/embedding/extract_para.py @@ -1,5 +1,5 @@ #!/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/embedding/paraconvert.py b/demo/model_zoo/embedding/paraconvert.py index 54155eff8e..ce7a70efc4 100755 --- a/demo/model_zoo/embedding/paraconvert.py +++ b/demo/model_zoo/embedding/paraconvert.py @@ -1,5 +1,5 @@ #!/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/embedding/pre_DictAndModel.sh b/demo/model_zoo/embedding/pre_DictAndModel.sh index 6d647f5dd9..f97ef26107 100755 --- a/demo/model_zoo/embedding/pre_DictAndModel.sh +++ b/demo/model_zoo/embedding/pre_DictAndModel.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/classify.py b/demo/model_zoo/resnet/classify.py index 7855126edc..4631816c43 100755 --- a/demo/model_zoo/resnet/classify.py +++ b/demo/model_zoo/resnet/classify.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/example/__init__.py b/demo/model_zoo/resnet/example/__init__.py index c90af2ee00..f662d68263 100644 --- a/demo/model_zoo/resnet/example/__init__.py +++ b/demo/model_zoo/resnet/example/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/example/image_list_provider.py b/demo/model_zoo/resnet/example/image_list_provider.py index 9e415f76a5..2cd8eb8bf8 100644 --- a/demo/model_zoo/resnet/example/image_list_provider.py +++ b/demo/model_zoo/resnet/example/image_list_provider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/extract_fea_c++.sh b/demo/model_zoo/resnet/extract_fea_c++.sh index c7f9aea9a5..5447aa92df 100755 --- a/demo/model_zoo/resnet/extract_fea_c++.sh +++ b/demo/model_zoo/resnet/extract_fea_c++.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/extract_fea_py.sh b/demo/model_zoo/resnet/extract_fea_py.sh index a70cef9a87..2e87152f7f 100755 --- a/demo/model_zoo/resnet/extract_fea_py.sh +++ b/demo/model_zoo/resnet/extract_fea_py.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/get_model.sh b/demo/model_zoo/resnet/get_model.sh index 133d08fca4..b33d8178ab 100755 --- a/demo/model_zoo/resnet/get_model.sh +++ b/demo/model_zoo/resnet/get_model.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/load_feature.py b/demo/model_zoo/resnet/load_feature.py index b0948b75fd..5d3d0c0d30 100644 --- a/demo/model_zoo/resnet/load_feature.py +++ b/demo/model_zoo/resnet/load_feature.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/net_diagram.sh b/demo/model_zoo/resnet/net_diagram.sh index a21ab4345b..1b06ffa44e 100755 --- a/demo/model_zoo/resnet/net_diagram.sh +++ b/demo/model_zoo/resnet/net_diagram.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/predict.sh b/demo/model_zoo/resnet/predict.sh index 55cf16e34a..2b67b17c48 100755 --- a/demo/model_zoo/resnet/predict.sh +++ b/demo/model_zoo/resnet/predict.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/model_zoo/resnet/resnet.py b/demo/model_zoo/resnet/resnet.py index 015b74cd48..6fdd97fefc 100644 --- a/demo/model_zoo/resnet/resnet.py +++ b/demo/model_zoo/resnet/resnet.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/api_train.py b/demo/quick_start/api_train.py index 66cbb85648..5699789daa 100644 --- a/demo/quick_start/api_train.py +++ b/demo/quick_start/api_train.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/api_train.sh b/demo/quick_start/api_train.sh index 40e9d0a09a..9b2a4e2f22 100755 --- a/demo/quick_start/api_train.sh +++ b/demo/quick_start/api_train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/data/get_data.sh b/demo/quick_start/data/get_data.sh index 952de3f3c8..a09a18f919 100755 --- a/demo/quick_start/data/get_data.sh +++ b/demo/quick_start/data/get_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/data/proc_from_raw_data/get_data.sh b/demo/quick_start/data/proc_from_raw_data/get_data.sh index cd85e26842..d976eaebfa 100755 --- a/demo/quick_start/data/proc_from_raw_data/get_data.sh +++ b/demo/quick_start/data/proc_from_raw_data/get_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/data/proc_from_raw_data/preprocess.py b/demo/quick_start/data/proc_from_raw_data/preprocess.py index 56c2c5f16c..72bd95f21d 100755 --- a/demo/quick_start/data/proc_from_raw_data/preprocess.py +++ b/demo/quick_start/data/proc_from_raw_data/preprocess.py @@ -1,6 +1,6 @@ # -*- coding: UTF-8 -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/dataprovider_bow.py b/demo/quick_start/dataprovider_bow.py index a5156a2d40..8e651d77bf 100644 --- a/demo/quick_start/dataprovider_bow.py +++ b/demo/quick_start/dataprovider_bow.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/dataprovider_emb.py b/demo/quick_start/dataprovider_emb.py index 286f3f5c82..b010253a8a 100755 --- a/demo/quick_start/dataprovider_emb.py +++ b/demo/quick_start/dataprovider_emb.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/predict.sh b/demo/quick_start/predict.sh index b1e5e44f0b..f02e5038e9 100755 --- a/demo/quick_start/predict.sh +++ b/demo/quick_start/predict.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/train.sh b/demo/quick_start/train.sh index b3c471608c..e3595fce75 100755 --- a/demo/quick_start/train.sh +++ b/demo/quick_start/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.bidi-lstm.py b/demo/quick_start/trainer_config.bidi-lstm.py index 51deaf31f9..ca1d1f8d09 100644 --- a/demo/quick_start/trainer_config.bidi-lstm.py +++ b/demo/quick_start/trainer_config.bidi-lstm.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.cnn.py b/demo/quick_start/trainer_config.cnn.py index 388efa75f9..f8c3d511f3 100644 --- a/demo/quick_start/trainer_config.cnn.py +++ b/demo/quick_start/trainer_config.cnn.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.db-lstm.py b/demo/quick_start/trainer_config.db-lstm.py index 02bc898d88..fba802b460 100644 --- a/demo/quick_start/trainer_config.db-lstm.py +++ b/demo/quick_start/trainer_config.db-lstm.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.emb.py b/demo/quick_start/trainer_config.emb.py index 8fd18a7aac..7410397ef6 100644 --- a/demo/quick_start/trainer_config.emb.py +++ b/demo/quick_start/trainer_config.emb.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.lr.py b/demo/quick_start/trainer_config.lr.py index b9c9441baa..e5105aa895 100644 --- a/demo/quick_start/trainer_config.lr.py +++ b/demo/quick_start/trainer_config.lr.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.lstm.py b/demo/quick_start/trainer_config.lstm.py index 8821e02d9b..43b4ddac2d 100644 --- a/demo/quick_start/trainer_config.lstm.py +++ b/demo/quick_start/trainer_config.lstm.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/quick_start/trainer_config.resnet-lstm.py b/demo/quick_start/trainer_config.resnet-lstm.py index 91e1581c38..5bed925d84 100644 --- a/demo/quick_start/trainer_config.resnet-lstm.py +++ b/demo/quick_start/trainer_config.resnet-lstm.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/common_utils.py b/demo/recommendation/common_utils.py index 613e36b496..d4fbdad1d7 100755 --- a/demo/recommendation/common_utils.py +++ b/demo/recommendation/common_utils.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/data/config_generator.py b/demo/recommendation/data/config_generator.py index fa60545830..4ca496a252 100644 --- a/demo/recommendation/data/config_generator.py +++ b/demo/recommendation/data/config_generator.py @@ -1,5 +1,5 @@ #!/bin/env python2 -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/data/meta_generator.py b/demo/recommendation/data/meta_generator.py index 593c863670..38e4679d26 100644 --- a/demo/recommendation/data/meta_generator.py +++ b/demo/recommendation/data/meta_generator.py @@ -1,5 +1,5 @@ #!/bin/env python2 -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/data/ml_data.sh b/demo/recommendation/data/ml_data.sh index 408a8723e0..2268d87638 100755 --- a/demo/recommendation/data/ml_data.sh +++ b/demo/recommendation/data/ml_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/data/split.py b/demo/recommendation/data/split.py index 8dd0cbd32a..be6869c22f 100644 --- a/demo/recommendation/data/split.py +++ b/demo/recommendation/data/split.py @@ -1,5 +1,5 @@ #!/bin/env python2 -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/dataprovider.py b/demo/recommendation/dataprovider.py index ff3932be03..80c62d7561 100755 --- a/demo/recommendation/dataprovider.py +++ b/demo/recommendation/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/evaluate.sh b/demo/recommendation/evaluate.sh index 38c1562c63..02b2857de0 100755 --- a/demo/recommendation/evaluate.sh +++ b/demo/recommendation/evaluate.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/prediction.py b/demo/recommendation/prediction.py index e2a202cfd1..191120188e 100755 --- a/demo/recommendation/prediction.py +++ b/demo/recommendation/prediction.py @@ -1,5 +1,5 @@ #!/bin/env python2 -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/preprocess.sh b/demo/recommendation/preprocess.sh index e181d0be45..e121e47019 100755 --- a/demo/recommendation/preprocess.sh +++ b/demo/recommendation/preprocess.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/run.sh b/demo/recommendation/run.sh index 846b59cec9..e341d1cc7a 100755 --- a/demo/recommendation/run.sh +++ b/demo/recommendation/run.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/recommendation/trainer_config.py b/demo/recommendation/trainer_config.py index cec340b0b6..aabcd33525 100755 --- a/demo/recommendation/trainer_config.py +++ b/demo/recommendation/trainer_config.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/data/extract_dict_feature.py b/demo/semantic_role_labeling/data/extract_dict_feature.py index daca5f01cf..123df022f5 100644 --- a/demo/semantic_role_labeling/data/extract_dict_feature.py +++ b/demo/semantic_role_labeling/data/extract_dict_feature.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/data/extract_pairs.py b/demo/semantic_role_labeling/data/extract_pairs.py index 86ab00ce41..2d0d535c53 100644 --- a/demo/semantic_role_labeling/data/extract_pairs.py +++ b/demo/semantic_role_labeling/data/extract_pairs.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/data/get_data.sh b/demo/semantic_role_labeling/data/get_data.sh index 99487e0d9a..a0ef26a13b 100644 --- a/demo/semantic_role_labeling/data/get_data.sh +++ b/demo/semantic_role_labeling/data/get_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/dataprovider.py b/demo/semantic_role_labeling/dataprovider.py index 2c8e134627..d12f10bfcb 100644 --- a/demo/semantic_role_labeling/dataprovider.py +++ b/demo/semantic_role_labeling/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/db_lstm.py b/demo/semantic_role_labeling/db_lstm.py index 54ceff0e72..75946bd72e 100644 --- a/demo/semantic_role_labeling/db_lstm.py +++ b/demo/semantic_role_labeling/db_lstm.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/predict.py b/demo/semantic_role_labeling/predict.py index a7f1e8f81f..15145fafce 100644 --- a/demo/semantic_role_labeling/predict.py +++ b/demo/semantic_role_labeling/predict.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/predict.sh b/demo/semantic_role_labeling/predict.sh index 88ab5898f7..873aad670d 100755 --- a/demo/semantic_role_labeling/predict.sh +++ b/demo/semantic_role_labeling/predict.sh @@ -1,6 +1,6 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/test.sh b/demo/semantic_role_labeling/test.sh index f9e1bdcd4c..11d9d6a19c 100755 --- a/demo/semantic_role_labeling/test.sh +++ b/demo/semantic_role_labeling/test.sh @@ -1,6 +1,6 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/semantic_role_labeling/train.sh b/demo/semantic_role_labeling/train.sh index 420768bb2b..9354e72f46 100755 --- a/demo/semantic_role_labeling/train.sh +++ b/demo/semantic_role_labeling/train.sh @@ -1,6 +1,6 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/data/get_imdb.sh b/demo/sentiment/data/get_imdb.sh index 28fa86232d..7600af6fbb 100755 --- a/demo/sentiment/data/get_imdb.sh +++ b/demo/sentiment/data/get_imdb.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/dataprovider.py b/demo/sentiment/dataprovider.py index 53e3d1d20d..00f72cecac 100755 --- a/demo/sentiment/dataprovider.py +++ b/demo/sentiment/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/predict.py b/demo/sentiment/predict.py index bc0f6f3126..00239c6009 100755 --- a/demo/sentiment/predict.py +++ b/demo/sentiment/predict.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/predict.sh b/demo/sentiment/predict.sh index 053f23e491..a889dfe3ec 100755 --- a/demo/sentiment/predict.sh +++ b/demo/sentiment/predict.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/preprocess.py b/demo/sentiment/preprocess.py index 7146e95d75..29b3682b74 100755 --- a/demo/sentiment/preprocess.py +++ b/demo/sentiment/preprocess.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/preprocess.sh b/demo/sentiment/preprocess.sh index 5f5c78e222..19ec34d4f0 100755 --- a/demo/sentiment/preprocess.sh +++ b/demo/sentiment/preprocess.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/sentiment_net.py b/demo/sentiment/sentiment_net.py index ff6a3624a4..a01577ca5a 100644 --- a/demo/sentiment/sentiment_net.py +++ b/demo/sentiment/sentiment_net.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/test.sh b/demo/sentiment/test.sh index c8b12a0e89..8af827c338 100755 --- a/demo/sentiment/test.sh +++ b/demo/sentiment/test.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/train.sh b/demo/sentiment/train.sh index f44a9a53f2..5ce8bf4b99 100755 --- a/demo/sentiment/train.sh +++ b/demo/sentiment/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sentiment/trainer_config.py b/demo/sentiment/trainer_config.py index 114a9138eb..2defecd178 100644 --- a/demo/sentiment/trainer_config.py +++ b/demo/sentiment/trainer_config.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/data/paraphrase_data.sh b/demo/seqToseq/data/paraphrase_data.sh index 1b3f1d45e1..e6497c9128 100755 --- a/demo/seqToseq/data/paraphrase_data.sh +++ b/demo/seqToseq/data/paraphrase_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/data/paraphrase_model.sh b/demo/seqToseq/data/paraphrase_model.sh index 041f69cf46..d0e7f214a3 100755 --- a/demo/seqToseq/data/paraphrase_model.sh +++ b/demo/seqToseq/data/paraphrase_model.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/data/wmt14_data.sh b/demo/seqToseq/data/wmt14_data.sh index 6c360b2060..43f67168d2 100755 --- a/demo/seqToseq/data/wmt14_data.sh +++ b/demo/seqToseq/data/wmt14_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/data/wmt14_model.sh b/demo/seqToseq/data/wmt14_model.sh index d6e7a73264..c4b55b90a3 100755 --- a/demo/seqToseq/data/wmt14_model.sh +++ b/demo/seqToseq/data/wmt14_model.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/dataprovider.py b/demo/seqToseq/dataprovider.py index 127c3672c7..c2b49804be 100755 --- a/demo/seqToseq/dataprovider.py +++ b/demo/seqToseq/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/paraphrase/train.conf b/demo/seqToseq/paraphrase/train.conf index 748920e2c7..be79c5e771 100644 --- a/demo/seqToseq/paraphrase/train.conf +++ b/demo/seqToseq/paraphrase/train.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/paraphrase/train.sh b/demo/seqToseq/paraphrase/train.sh index 2aa7b84060..33a42f6eff 100755 --- a/demo/seqToseq/paraphrase/train.sh +++ b/demo/seqToseq/paraphrase/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/preprocess.py b/demo/seqToseq/preprocess.py index bd1c51b151..03f371331a 100755 --- a/demo/seqToseq/preprocess.py +++ b/demo/seqToseq/preprocess.py @@ -1,5 +1,5 @@ #!/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/seqToseq_net.py b/demo/seqToseq/seqToseq_net.py index fc9db05ba7..e523a34d5a 100644 --- a/demo/seqToseq/seqToseq_net.py +++ b/demo/seqToseq/seqToseq_net.py @@ -1,6 +1,6 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/eval_bleu.sh b/demo/seqToseq/translation/eval_bleu.sh index ef0ede717a..54c2ed237e 100755 --- a/demo/seqToseq/translation/eval_bleu.sh +++ b/demo/seqToseq/translation/eval_bleu.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/gen.conf b/demo/seqToseq/translation/gen.conf index 63c5c2f9a6..e9bea4e455 100644 --- a/demo/seqToseq/translation/gen.conf +++ b/demo/seqToseq/translation/gen.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/gen.sh b/demo/seqToseq/translation/gen.sh index ad977c05ff..a700ae2134 100755 --- a/demo/seqToseq/translation/gen.sh +++ b/demo/seqToseq/translation/gen.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/moses_bleu.sh b/demo/seqToseq/translation/moses_bleu.sh index bfaba40b26..2f230d7f4c 100755 --- a/demo/seqToseq/translation/moses_bleu.sh +++ b/demo/seqToseq/translation/moses_bleu.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/train.conf b/demo/seqToseq/translation/train.conf index cf1bde15c4..72b7ccdbb9 100644 --- a/demo/seqToseq/translation/train.conf +++ b/demo/seqToseq/translation/train.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/seqToseq/translation/train.sh b/demo/seqToseq/translation/train.sh index 976b5ba3b0..bdece693e5 100755 --- a/demo/seqToseq/translation/train.sh +++ b/demo/seqToseq/translation/train.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sequence_tagging/data/get_data.sh b/demo/sequence_tagging/data/get_data.sh index e579d6c46c..0cdb394035 100755 --- a/demo/sequence_tagging/data/get_data.sh +++ b/demo/sequence_tagging/data/get_data.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sequence_tagging/dataprovider.py b/demo/sequence_tagging/dataprovider.py index 37dcb7aa17..bb4b4465bc 100644 --- a/demo/sequence_tagging/dataprovider.py +++ b/demo/sequence_tagging/dataprovider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sequence_tagging/linear_crf.py b/demo/sequence_tagging/linear_crf.py index 64895742e1..736b580bb8 100644 --- a/demo/sequence_tagging/linear_crf.py +++ b/demo/sequence_tagging/linear_crf.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/demo/sequence_tagging/rnn_crf.py b/demo/sequence_tagging/rnn_crf.py index 90d4bbdddf..ad1e7b68e7 100644 --- a/demo/sequence_tagging/rnn_crf.py +++ b/demo/sequence_tagging/rnn_crf.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/doc/api/predict/predict_sample.py b/doc/api/predict/predict_sample.py index 63e8b36d26..51349250e8 100644 --- a/doc/api/predict/predict_sample.py +++ b/doc/api/predict/predict_sample.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/doc_cn/cluster/k8s/start_paddle.py b/doc_cn/cluster/k8s/start_paddle.py index bc0112a77f..6a46161410 100755 --- a/doc_cn/cluster/k8s/start_paddle.py +++ b/doc_cn/cluster/k8s/start_paddle.py @@ -1,5 +1,5 @@ #!/usr/bin/python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/.common_test_util.sh b/paddle/.common_test_util.sh index dc15250615..8d024bc7d0 100644 --- a/paddle/.common_test_util.sh +++ b/paddle/.common_test_util.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/.set_port.sh b/paddle/.set_port.sh index 33596fac60..617ac79a24 100755 --- a/paddle/.set_port.sh +++ b/paddle/.set_port.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/.set_python_path.sh b/paddle/.set_python_path.sh index 657fdf65e9..fa7baccc86 100755 --- a/paddle/.set_python_path.sh +++ b/paddle/.set_python_path.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/Arguments.cpp b/paddle/api/Arguments.cpp index bd1fdffe89..0cafbd896e 100644 --- a/paddle/api/Arguments.cpp +++ b/paddle/api/Arguments.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/ConfigParser.cpp b/paddle/api/ConfigParser.cpp index bc40d871d1..2f45173bfd 100644 --- a/paddle/api/ConfigParser.cpp +++ b/paddle/api/ConfigParser.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/GradientMachine.cpp b/paddle/api/GradientMachine.cpp index 9a4846d809..c1b546dbcb 100644 --- a/paddle/api/GradientMachine.cpp +++ b/paddle/api/GradientMachine.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Internal.h b/paddle/api/Internal.h index 66a13bc603..4a07880d80 100644 --- a/paddle/api/Internal.h +++ b/paddle/api/Internal.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Matrix.cpp b/paddle/api/Matrix.cpp index f257ee65aa..d4c00e7093 100644 --- a/paddle/api/Matrix.cpp +++ b/paddle/api/Matrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/PaddleAPI.h b/paddle/api/PaddleAPI.h index a125934fc1..f3c80e3b06 100644 --- a/paddle/api/PaddleAPI.h +++ b/paddle/api/PaddleAPI.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/PaddleAPIPrivate.h b/paddle/api/PaddleAPIPrivate.h index 5ffeff6a97..d2b56fc41c 100644 --- a/paddle/api/PaddleAPIPrivate.h +++ b/paddle/api/PaddleAPIPrivate.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Parameter.cpp b/paddle/api/Parameter.cpp index 9c30ef6ff4..742ad0679c 100644 --- a/paddle/api/Parameter.cpp +++ b/paddle/api/Parameter.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/ParameterOptimizer.cpp b/paddle/api/ParameterOptimizer.cpp index 21d031e4bc..606dccd5ac 100644 --- a/paddle/api/ParameterOptimizer.cpp +++ b/paddle/api/ParameterOptimizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/SequenceGenerator.cpp b/paddle/api/SequenceGenerator.cpp index d51be78d45..5c65b34f23 100644 --- a/paddle/api/SequenceGenerator.cpp +++ b/paddle/api/SequenceGenerator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Trainer.cpp b/paddle/api/Trainer.cpp index 7a6aa69fb6..9aeb874bdc 100644 --- a/paddle/api/Trainer.cpp +++ b/paddle/api/Trainer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Util.cpp b/paddle/api/Util.cpp index 1bba1df2e1..0c9c048099 100644 --- a/paddle/api/Util.cpp +++ b/paddle/api/Util.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/Vector.cpp b/paddle/api/Vector.cpp index 74c9ff8dc7..4f3ab7de60 100644 --- a/paddle/api/Vector.cpp +++ b/paddle/api/Vector.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/api/__init__.py b/paddle/api/__init__.py index c90af2ee00..f662d68263 100644 --- a/paddle/api/__init__.py +++ b/paddle/api/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/paddle_ld_flags.py b/paddle/api/paddle_ld_flags.py index ebe00798e8..85cc54700f 100644 --- a/paddle/api/paddle_ld_flags.py +++ b/paddle/api/paddle_ld_flags.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/run_tests.sh b/paddle/api/test/run_tests.sh index ff69c45264..2f12ba0264 100755 --- a/paddle/api/test/run_tests.sh +++ b/paddle/api/test/run_tests.sh @@ -1,5 +1,5 @@ #!/bin/bash -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testArguments.py b/paddle/api/test/testArguments.py index 70fb169fd5..8cabecd242 100644 --- a/paddle/api/test/testArguments.py +++ b/paddle/api/test/testArguments.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testGradientMachine.py b/paddle/api/test/testGradientMachine.py index e12613fbb8..b81eafa967 100644 --- a/paddle/api/test/testGradientMachine.py +++ b/paddle/api/test/testGradientMachine.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testMatrix.py b/paddle/api/test/testMatrix.py index 8b0da62692..f76f84d2e1 100644 --- a/paddle/api/test/testMatrix.py +++ b/paddle/api/test/testMatrix.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testTrain.py b/paddle/api/test/testTrain.py index a3ba4eaaa6..a90d15c272 100644 --- a/paddle/api/test/testTrain.py +++ b/paddle/api/test/testTrain.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testTrainer.py b/paddle/api/test/testTrainer.py index edd5a2da57..a76cbf02d8 100644 --- a/paddle/api/test/testTrainer.py +++ b/paddle/api/test/testTrainer.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/testVector.py b/paddle/api/test/testVector.py index 963359236d..525ed97edd 100644 --- a/paddle/api/test/testVector.py +++ b/paddle/api/test/testVector.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/api/test/util.py b/paddle/api/test/util.py index dbcdba5bf2..9f4631c53e 100644 --- a/paddle/api/test/util.py +++ b/paddle/api/test/util.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_activation_functions.h b/paddle/cuda/include/hl_activation_functions.h index 03e15b2223..cdb2dba06c 100644 --- a/paddle/cuda/include/hl_activation_functions.h +++ b/paddle/cuda/include/hl_activation_functions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_aggregate.h b/paddle/cuda/include/hl_aggregate.h index a6d9ff8483..d2189de689 100644 --- a/paddle/cuda/include/hl_aggregate.h +++ b/paddle/cuda/include/hl_aggregate.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_avx_functions.h b/paddle/cuda/include/hl_avx_functions.h index ed339e312a..35f4eabb4c 100644 --- a/paddle/cuda/include/hl_avx_functions.h +++ b/paddle/cuda/include/hl_avx_functions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_base.h b/paddle/cuda/include/hl_base.h index a076952467..0b9dfc6117 100644 --- a/paddle/cuda/include/hl_base.h +++ b/paddle/cuda/include/hl_base.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_batch_transpose.h b/paddle/cuda/include/hl_batch_transpose.h index f3630e9762..e2e958cd67 100644 --- a/paddle/cuda/include/hl_batch_transpose.h +++ b/paddle/cuda/include/hl_batch_transpose.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cnn.h b/paddle/cuda/include/hl_cnn.h index cffaac634f..06ee3b3654 100644 --- a/paddle/cuda/include/hl_cnn.h +++ b/paddle/cuda/include/hl_cnn.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cpu_gru.cuh b/paddle/cuda/include/hl_cpu_gru.cuh index d39cf67448..c0a37ced2a 100644 --- a/paddle/cuda/include/hl_cpu_gru.cuh +++ b/paddle/cuda/include/hl_cpu_gru.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cpu_lstm.cuh b/paddle/cuda/include/hl_cpu_lstm.cuh index 65a174d85b..0e412fcdf5 100644 --- a/paddle/cuda/include/hl_cpu_lstm.cuh +++ b/paddle/cuda/include/hl_cpu_lstm.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cpu_matrix_kernel.cuh b/paddle/cuda/include/hl_cpu_matrix_kernel.cuh index 239a241991..f35bfbc5c8 100644 --- a/paddle/cuda/include/hl_cpu_matrix_kernel.cuh +++ b/paddle/cuda/include/hl_cpu_matrix_kernel.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cuda.h b/paddle/cuda/include/hl_cuda.h index 2c7d665101..5383c1130b 100644 --- a/paddle/cuda/include/hl_cuda.h +++ b/paddle/cuda/include/hl_cuda.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cuda.ph b/paddle/cuda/include/hl_cuda.ph index 9e0537aaf1..701916b279 100644 --- a/paddle/cuda/include/hl_cuda.ph +++ b/paddle/cuda/include/hl_cuda.ph @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cuda_cublas.h b/paddle/cuda/include/hl_cuda_cublas.h index db8c03c2c0..e206e42b2a 100644 --- a/paddle/cuda/include/hl_cuda_cublas.h +++ b/paddle/cuda/include/hl_cuda_cublas.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cuda_cudnn.h b/paddle/cuda/include/hl_cuda_cudnn.h index 3a2f916210..db18e4912b 100644 --- a/paddle/cuda/include/hl_cuda_cudnn.h +++ b/paddle/cuda/include/hl_cuda_cudnn.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_cuda_cudnn.ph b/paddle/cuda/include/hl_cuda_cudnn.ph index c0e82abe17..61378937ce 100644 --- a/paddle/cuda/include/hl_cuda_cudnn.ph +++ b/paddle/cuda/include/hl_cuda_cudnn.ph @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_device_functions.cuh b/paddle/cuda/include/hl_device_functions.cuh index 159c26f443..e0b5632f23 100755 --- a/paddle/cuda/include/hl_device_functions.cuh +++ b/paddle/cuda/include/hl_device_functions.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_dso_loader.h b/paddle/cuda/include/hl_dso_loader.h index e5d3d40311..9ddf0e61ee 100644 --- a/paddle/cuda/include/hl_dso_loader.h +++ b/paddle/cuda/include/hl_dso_loader.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_functions.h b/paddle/cuda/include/hl_functions.h index 91ce9a0678..0d7e80a855 100644 --- a/paddle/cuda/include/hl_functions.h +++ b/paddle/cuda/include/hl_functions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gpu.h b/paddle/cuda/include/hl_gpu.h index 6dd6d13212..aad0450c8c 100644 --- a/paddle/cuda/include/hl_gpu.h +++ b/paddle/cuda/include/hl_gpu.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gpu_functions.cuh b/paddle/cuda/include/hl_gpu_functions.cuh index a2c5ebd18a..8e64cbe360 100644 --- a/paddle/cuda/include/hl_gpu_functions.cuh +++ b/paddle/cuda/include/hl_gpu_functions.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gpu_gru.cuh b/paddle/cuda/include/hl_gpu_gru.cuh index 3e0cfdbe4f..6668e135d2 100644 --- a/paddle/cuda/include/hl_gpu_gru.cuh +++ b/paddle/cuda/include/hl_gpu_gru.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gpu_lstm.cuh b/paddle/cuda/include/hl_gpu_lstm.cuh index 07806e11c1..5dceba2f5b 100644 --- a/paddle/cuda/include/hl_gpu_lstm.cuh +++ b/paddle/cuda/include/hl_gpu_lstm.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gpu_matrix_kernel.cuh b/paddle/cuda/include/hl_gpu_matrix_kernel.cuh index 201c5c25f1..9bbdf5fa72 100644 --- a/paddle/cuda/include/hl_gpu_matrix_kernel.cuh +++ b/paddle/cuda/include/hl_gpu_matrix_kernel.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_gru_ops.cuh b/paddle/cuda/include/hl_gru_ops.cuh index 3c137d8d44..45f66ad533 100644 --- a/paddle/cuda/include/hl_gru_ops.cuh +++ b/paddle/cuda/include/hl_gru_ops.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_lstm.h b/paddle/cuda/include/hl_lstm.h index 7e527a7902..857756e5cd 100644 --- a/paddle/cuda/include/hl_lstm.h +++ b/paddle/cuda/include/hl_lstm.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_lstm_ops.cuh b/paddle/cuda/include/hl_lstm_ops.cuh index a5ea018dbc..2601060cc2 100644 --- a/paddle/cuda/include/hl_lstm_ops.cuh +++ b/paddle/cuda/include/hl_lstm_ops.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix.h b/paddle/cuda/include/hl_matrix.h index 96648661e3..abd5eb3a0c 100644 --- a/paddle/cuda/include/hl_matrix.h +++ b/paddle/cuda/include/hl_matrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix_apply.cuh b/paddle/cuda/include/hl_matrix_apply.cuh index 927212c83d..b10d177b97 100644 --- a/paddle/cuda/include/hl_matrix_apply.cuh +++ b/paddle/cuda/include/hl_matrix_apply.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix_base.cuh b/paddle/cuda/include/hl_matrix_base.cuh index a3645ef51e..db35ee2037 100644 --- a/paddle/cuda/include/hl_matrix_base.cuh +++ b/paddle/cuda/include/hl_matrix_base.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix_base_sse.cuh b/paddle/cuda/include/hl_matrix_base_sse.cuh index dd55b84884..db6c9cca03 100644 --- a/paddle/cuda/include/hl_matrix_base_sse.cuh +++ b/paddle/cuda/include/hl_matrix_base_sse.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix_ops.cuh b/paddle/cuda/include/hl_matrix_ops.cuh index 3e5e1bc701..fc29201357 100644 --- a/paddle/cuda/include/hl_matrix_ops.cuh +++ b/paddle/cuda/include/hl_matrix_ops.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_matrix_type.cuh b/paddle/cuda/include/hl_matrix_type.cuh index 060be07364..59213eee75 100644 --- a/paddle/cuda/include/hl_matrix_type.cuh +++ b/paddle/cuda/include/hl_matrix_type.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_perturbation_util.cuh b/paddle/cuda/include/hl_perturbation_util.cuh index 90fc1cb060..93b81bf035 100644 --- a/paddle/cuda/include/hl_perturbation_util.cuh +++ b/paddle/cuda/include/hl_perturbation_util.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_recurrent_apply.cuh b/paddle/cuda/include/hl_recurrent_apply.cuh index 0ccbf01f1c..113446cf75 100644 --- a/paddle/cuda/include/hl_recurrent_apply.cuh +++ b/paddle/cuda/include/hl_recurrent_apply.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_sequence.h b/paddle/cuda/include/hl_sequence.h index b98d7bdeaf..9bcd25b062 100644 --- a/paddle/cuda/include/hl_sequence.h +++ b/paddle/cuda/include/hl_sequence.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_sparse.h b/paddle/cuda/include/hl_sparse.h index c4e0be23e2..67fe701c10 100644 --- a/paddle/cuda/include/hl_sparse.h +++ b/paddle/cuda/include/hl_sparse.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_sparse.ph b/paddle/cuda/include/hl_sparse.ph index d3bc73c80d..13bba17811 100644 --- a/paddle/cuda/include/hl_sparse.ph +++ b/paddle/cuda/include/hl_sparse.ph @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_sse_matrix_kernel.cuh b/paddle/cuda/include/hl_sse_matrix_kernel.cuh index 45db2f313e..9e50580669 100644 --- a/paddle/cuda/include/hl_sse_matrix_kernel.cuh +++ b/paddle/cuda/include/hl_sse_matrix_kernel.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_table_apply.h b/paddle/cuda/include/hl_table_apply.h index b4ac83a66a..2170b97f4d 100644 --- a/paddle/cuda/include/hl_table_apply.h +++ b/paddle/cuda/include/hl_table_apply.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_tensor_ops.h b/paddle/cuda/include/hl_tensor_ops.h index cc95620e37..7945b98201 100644 --- a/paddle/cuda/include/hl_tensor_ops.h +++ b/paddle/cuda/include/hl_tensor_ops.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_thread.ph b/paddle/cuda/include/hl_thread.ph index 0cfc459936..a3830ff8d8 100644 --- a/paddle/cuda/include/hl_thread.ph +++ b/paddle/cuda/include/hl_thread.ph @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_time.h b/paddle/cuda/include/hl_time.h index b0a88c66a1..f214b055f9 100644 --- a/paddle/cuda/include/hl_time.h +++ b/paddle/cuda/include/hl_time.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_top_k.h b/paddle/cuda/include/hl_top_k.h index e8cfebbf6a..77949ed295 100644 --- a/paddle/cuda/include/hl_top_k.h +++ b/paddle/cuda/include/hl_top_k.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/hl_warpctc_wrap.h b/paddle/cuda/include/hl_warpctc_wrap.h index dc50cf9d20..79bf6c3db7 100644 --- a/paddle/cuda/include/hl_warpctc_wrap.h +++ b/paddle/cuda/include/hl_warpctc_wrap.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_aggregate_stub.h b/paddle/cuda/include/stub/hl_aggregate_stub.h index bb53fc581e..bbfa9b8fad 100644 --- a/paddle/cuda/include/stub/hl_aggregate_stub.h +++ b/paddle/cuda/include/stub/hl_aggregate_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_cnn_stub.h b/paddle/cuda/include/stub/hl_cnn_stub.h index 2f73b9671e..52c9787352 100644 --- a/paddle/cuda/include/stub/hl_cnn_stub.h +++ b/paddle/cuda/include/stub/hl_cnn_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_cuda_cublas_stub.h b/paddle/cuda/include/stub/hl_cuda_cublas_stub.h index 85f7c390c4..e86fd853f4 100644 --- a/paddle/cuda/include/stub/hl_cuda_cublas_stub.h +++ b/paddle/cuda/include/stub/hl_cuda_cublas_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h b/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h index 3beb0e5b51..abd0d6b099 100644 --- a/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h +++ b/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_cuda_stub.h b/paddle/cuda/include/stub/hl_cuda_stub.h index 24923a0d4a..5246a8d5a4 100644 --- a/paddle/cuda/include/stub/hl_cuda_stub.h +++ b/paddle/cuda/include/stub/hl_cuda_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_lstm_stub.h b/paddle/cuda/include/stub/hl_lstm_stub.h index 7ccda032d2..246ba79f63 100644 --- a/paddle/cuda/include/stub/hl_lstm_stub.h +++ b/paddle/cuda/include/stub/hl_lstm_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_matrix_stub.h b/paddle/cuda/include/stub/hl_matrix_stub.h index 1bd78d23fb..0b669f6735 100644 --- a/paddle/cuda/include/stub/hl_matrix_stub.h +++ b/paddle/cuda/include/stub/hl_matrix_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_sequence_stub.h b/paddle/cuda/include/stub/hl_sequence_stub.h index 3343463a8d..d6b07556f8 100644 --- a/paddle/cuda/include/stub/hl_sequence_stub.h +++ b/paddle/cuda/include/stub/hl_sequence_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/include/stub/hl_sparse_stub.h b/paddle/cuda/include/stub/hl_sparse_stub.h index d47bdd2c47..bd17461d88 100644 --- a/paddle/cuda/include/stub/hl_sparse_stub.h +++ b/paddle/cuda/include/stub/hl_sparse_stub.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_avx_functions.cc b/paddle/cuda/src/hl_avx_functions.cc index c1e0c7f9d9..9066475876 100644 --- a/paddle/cuda/src/hl_avx_functions.cc +++ b/paddle/cuda/src/hl_avx_functions.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_batch_transpose.cu b/paddle/cuda/src/hl_batch_transpose.cu index 00fd18e7f3..f047403da1 100644 --- a/paddle/cuda/src/hl_batch_transpose.cu +++ b/paddle/cuda/src/hl_batch_transpose.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cpu_functions.cc b/paddle/cuda/src/hl_cpu_functions.cc index af00f352e5..c2117a7315 100644 --- a/paddle/cuda/src/hl_cpu_functions.cc +++ b/paddle/cuda/src/hl_cpu_functions.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_aggregate.cu b/paddle/cuda/src/hl_cuda_aggregate.cu index 4eb775eb79..97034a9177 100644 --- a/paddle/cuda/src/hl_cuda_aggregate.cu +++ b/paddle/cuda/src/hl_cuda_aggregate.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_cnn.cu b/paddle/cuda/src/hl_cuda_cnn.cu index 7f2f6897b4..0992286f36 100644 --- a/paddle/cuda/src/hl_cuda_cnn.cu +++ b/paddle/cuda/src/hl_cuda_cnn.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_cublas.cc b/paddle/cuda/src/hl_cuda_cublas.cc index e8ba232d44..7cede8c63c 100644 --- a/paddle/cuda/src/hl_cuda_cublas.cc +++ b/paddle/cuda/src/hl_cuda_cublas.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_cudnn.cc b/paddle/cuda/src/hl_cuda_cudnn.cc index 9d4ff08a78..9c9b8906c2 100644 --- a/paddle/cuda/src/hl_cuda_cudnn.cc +++ b/paddle/cuda/src/hl_cuda_cudnn.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_device.cc b/paddle/cuda/src/hl_cuda_device.cc index 6b71a53848..d181448292 100644 --- a/paddle/cuda/src/hl_cuda_device.cc +++ b/paddle/cuda/src/hl_cuda_device.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_lstm.cu b/paddle/cuda/src/hl_cuda_lstm.cu index cf009620bf..b869d903ba 100644 --- a/paddle/cuda/src/hl_cuda_lstm.cu +++ b/paddle/cuda/src/hl_cuda_lstm.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_matrix.cu b/paddle/cuda/src/hl_cuda_matrix.cu index 0b7cd33756..2b4c6f7c39 100644 --- a/paddle/cuda/src/hl_cuda_matrix.cu +++ b/paddle/cuda/src/hl_cuda_matrix.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_sequence.cu b/paddle/cuda/src/hl_cuda_sequence.cu index e83a60ad72..4e33ac443c 100644 --- a/paddle/cuda/src/hl_cuda_sequence.cu +++ b/paddle/cuda/src/hl_cuda_sequence.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_sparse.cu b/paddle/cuda/src/hl_cuda_sparse.cu index 1687fcc221..ab9ab57c88 100644 --- a/paddle/cuda/src/hl_cuda_sparse.cu +++ b/paddle/cuda/src/hl_cuda_sparse.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cuda_sparse.cuh b/paddle/cuda/src/hl_cuda_sparse.cuh index 9cf2d5a843..72572756a6 100644 --- a/paddle/cuda/src/hl_cuda_sparse.cuh +++ b/paddle/cuda/src/hl_cuda_sparse.cuh @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_cudart_wrap.cc b/paddle/cuda/src/hl_cudart_wrap.cc index a95f5557af..a3ac750b53 100644 --- a/paddle/cuda/src/hl_cudart_wrap.cc +++ b/paddle/cuda/src/hl_cudart_wrap.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_dso_loader.cc b/paddle/cuda/src/hl_dso_loader.cc index ce19073626..f509b89243 100644 --- a/paddle/cuda/src/hl_dso_loader.cc +++ b/paddle/cuda/src/hl_dso_loader.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_math.cc b/paddle/cuda/src/hl_math.cc index f4bf888bab..3048693fb8 100644 --- a/paddle/cuda/src/hl_math.cc +++ b/paddle/cuda/src/hl_math.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_perturbation_util.cu b/paddle/cuda/src/hl_perturbation_util.cu index a10d06f8a9..2a945bcdb8 100644 --- a/paddle/cuda/src/hl_perturbation_util.cu +++ b/paddle/cuda/src/hl_perturbation_util.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_table_apply.cu b/paddle/cuda/src/hl_table_apply.cu index 52ee4610ed..61edbe3ccc 100644 --- a/paddle/cuda/src/hl_table_apply.cu +++ b/paddle/cuda/src/hl_table_apply.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_time.cc b/paddle/cuda/src/hl_time.cc index d52b2a1df0..3005065899 100644 --- a/paddle/cuda/src/hl_time.cc +++ b/paddle/cuda/src/hl_time.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_top_k.cu b/paddle/cuda/src/hl_top_k.cu index ed74787b61..f0ef0cc3c5 100644 --- a/paddle/cuda/src/hl_top_k.cu +++ b/paddle/cuda/src/hl_top_k.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/cuda/src/hl_warpctc_wrap.cc b/paddle/cuda/src/hl_warpctc_wrap.cc index 3d3bf46158..619b90120f 100644 --- a/paddle/cuda/src/hl_warpctc_wrap.cc +++ b/paddle/cuda/src/hl_warpctc_wrap.cc @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/activations/ActivationFunction.cpp b/paddle/gserver/activations/ActivationFunction.cpp index 220f220e0f..f1d09c568d 100644 --- a/paddle/gserver/activations/ActivationFunction.cpp +++ b/paddle/gserver/activations/ActivationFunction.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/activations/ActivationFunction.h b/paddle/gserver/activations/ActivationFunction.h index e9ed5c619a..601e3b6c0c 100644 --- a/paddle/gserver/activations/ActivationFunction.h +++ b/paddle/gserver/activations/ActivationFunction.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/DataProvider.cpp b/paddle/gserver/dataproviders/DataProvider.cpp index e6cc4a246a..55ca62543a 100644 --- a/paddle/gserver/dataproviders/DataProvider.cpp +++ b/paddle/gserver/dataproviders/DataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/DataProvider.h b/paddle/gserver/dataproviders/DataProvider.h index 8247693822..5b854936c6 100644 --- a/paddle/gserver/dataproviders/DataProvider.h +++ b/paddle/gserver/dataproviders/DataProvider.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/DataProviderGroup.h b/paddle/gserver/dataproviders/DataProviderGroup.h index 6c178e29ee..69ac2590b9 100644 --- a/paddle/gserver/dataproviders/DataProviderGroup.h +++ b/paddle/gserver/dataproviders/DataProviderGroup.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/MultiDataProvider.cpp b/paddle/gserver/dataproviders/MultiDataProvider.cpp index 51fb1f2666..e1fc4c9365 100644 --- a/paddle/gserver/dataproviders/MultiDataProvider.cpp +++ b/paddle/gserver/dataproviders/MultiDataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/MultiDataProvider.h b/paddle/gserver/dataproviders/MultiDataProvider.h index 876467c04f..4c8fb2cd0d 100644 --- a/paddle/gserver/dataproviders/MultiDataProvider.h +++ b/paddle/gserver/dataproviders/MultiDataProvider.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/ProtoDataProvider.cpp b/paddle/gserver/dataproviders/ProtoDataProvider.cpp index 0a7ff80246..6a0cb5ef63 100644 --- a/paddle/gserver/dataproviders/ProtoDataProvider.cpp +++ b/paddle/gserver/dataproviders/ProtoDataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/ProtoDataProvider.h b/paddle/gserver/dataproviders/ProtoDataProvider.h index ffdcc8fdc9..9ec5cb97c0 100644 --- a/paddle/gserver/dataproviders/ProtoDataProvider.h +++ b/paddle/gserver/dataproviders/ProtoDataProvider.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/ProtoReader.h b/paddle/gserver/dataproviders/ProtoReader.h index b8fca3cd7f..6708e7cde7 100644 --- a/paddle/gserver/dataproviders/ProtoReader.h +++ b/paddle/gserver/dataproviders/ProtoReader.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/PyDataProvider.cpp b/paddle/gserver/dataproviders/PyDataProvider.cpp index bee6ca14a2..f5dcbfcf34 100644 --- a/paddle/gserver/dataproviders/PyDataProvider.cpp +++ b/paddle/gserver/dataproviders/PyDataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/PyDataProvider.h b/paddle/gserver/dataproviders/PyDataProvider.h index 6bb7c831fd..1401c13a1e 100644 --- a/paddle/gserver/dataproviders/PyDataProvider.h +++ b/paddle/gserver/dataproviders/PyDataProvider.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/dataproviders/PyDataProvider2.cpp b/paddle/gserver/dataproviders/PyDataProvider2.cpp index 967fc9026a..8b04a03f6d 100644 --- a/paddle/gserver/dataproviders/PyDataProvider2.cpp +++ b/paddle/gserver/dataproviders/PyDataProvider2.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/evaluators/CTCErrorEvaluator.cpp b/paddle/gserver/evaluators/CTCErrorEvaluator.cpp index 8f7d2fb80e..05aa6c012a 100644 --- a/paddle/gserver/evaluators/CTCErrorEvaluator.cpp +++ b/paddle/gserver/evaluators/CTCErrorEvaluator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/evaluators/ChunkEvaluator.cpp b/paddle/gserver/evaluators/ChunkEvaluator.cpp index 923e77fc9d..3d8af5bcd4 100644 --- a/paddle/gserver/evaluators/ChunkEvaluator.cpp +++ b/paddle/gserver/evaluators/ChunkEvaluator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/evaluators/Evaluator.cpp b/paddle/gserver/evaluators/Evaluator.cpp index f5df2b18de..aa6dc7cb86 100644 --- a/paddle/gserver/evaluators/Evaluator.cpp +++ b/paddle/gserver/evaluators/Evaluator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/evaluators/Evaluator.h b/paddle/gserver/evaluators/Evaluator.h index 732abb6079..a26c650c38 100644 --- a/paddle/gserver/evaluators/Evaluator.h +++ b/paddle/gserver/evaluators/Evaluator.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/GradientMachine.cpp b/paddle/gserver/gradientmachines/GradientMachine.cpp index 3761fda5f3..6adee05dbe 100644 --- a/paddle/gserver/gradientmachines/GradientMachine.cpp +++ b/paddle/gserver/gradientmachines/GradientMachine.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/GradientMachine.h b/paddle/gserver/gradientmachines/GradientMachine.h index 27cdf7f789..f3e44a9e39 100644 --- a/paddle/gserver/gradientmachines/GradientMachine.h +++ b/paddle/gserver/gradientmachines/GradientMachine.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/GradientMachineMode.cpp b/paddle/gserver/gradientmachines/GradientMachineMode.cpp index 4a90a4a566..3583fb4de8 100644 --- a/paddle/gserver/gradientmachines/GradientMachineMode.cpp +++ b/paddle/gserver/gradientmachines/GradientMachineMode.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/GradientMachineMode.h b/paddle/gserver/gradientmachines/GradientMachineMode.h index f2f55a7067..7bc885fe99 100644 --- a/paddle/gserver/gradientmachines/GradientMachineMode.h +++ b/paddle/gserver/gradientmachines/GradientMachineMode.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/MultiGradientMachine.cpp b/paddle/gserver/gradientmachines/MultiGradientMachine.cpp index 148451f18d..a7324f5545 100644 --- a/paddle/gserver/gradientmachines/MultiGradientMachine.cpp +++ b/paddle/gserver/gradientmachines/MultiGradientMachine.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/MultiGradientMachine.h b/paddle/gserver/gradientmachines/MultiGradientMachine.h index 58c5486810..fe6d96e8ea 100644 --- a/paddle/gserver/gradientmachines/MultiGradientMachine.h +++ b/paddle/gserver/gradientmachines/MultiGradientMachine.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/MultiNetwork.cpp b/paddle/gserver/gradientmachines/MultiNetwork.cpp index e5be19cad6..61af82fcb7 100644 --- a/paddle/gserver/gradientmachines/MultiNetwork.cpp +++ b/paddle/gserver/gradientmachines/MultiNetwork.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/MultiNetwork.h b/paddle/gserver/gradientmachines/MultiNetwork.h index 779a2267f5..89fbf32b4f 100644 --- a/paddle/gserver/gradientmachines/MultiNetwork.h +++ b/paddle/gserver/gradientmachines/MultiNetwork.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/NeuralNetwork.cpp b/paddle/gserver/gradientmachines/NeuralNetwork.cpp index 9932ea655e..dbcb97b42b 100644 --- a/paddle/gserver/gradientmachines/NeuralNetwork.cpp +++ b/paddle/gserver/gradientmachines/NeuralNetwork.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/NeuralNetwork.h b/paddle/gserver/gradientmachines/NeuralNetwork.h index 55ef45c5ee..fd885b436a 100644 --- a/paddle/gserver/gradientmachines/NeuralNetwork.h +++ b/paddle/gserver/gradientmachines/NeuralNetwork.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp index 9dbf418c31..980a5851a2 100644 --- a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp +++ b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h index 71488bc3b7..934a7cfc7b 100644 --- a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h +++ b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp b/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp index 516b617576..4fb1a44ab7 100644 --- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp +++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/gradientmachines/RecurrentGradientMachine.h b/paddle/gserver/gradientmachines/RecurrentGradientMachine.h index cb74a67e52..369c8c3d98 100644 --- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.h +++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AddtoLayer.cpp b/paddle/gserver/layers/AddtoLayer.cpp index 8a9aecfa19..5338530113 100644 --- a/paddle/gserver/layers/AddtoLayer.cpp +++ b/paddle/gserver/layers/AddtoLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AddtoLayer.h b/paddle/gserver/layers/AddtoLayer.h index 883d186f3e..53d3f99cdd 100644 --- a/paddle/gserver/layers/AddtoLayer.h +++ b/paddle/gserver/layers/AddtoLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AgentLayer.cpp b/paddle/gserver/layers/AgentLayer.cpp index eb89281cb1..2d30029027 100644 --- a/paddle/gserver/layers/AgentLayer.cpp +++ b/paddle/gserver/layers/AgentLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AgentLayer.h b/paddle/gserver/layers/AgentLayer.h index 0186653c0f..41683ad671 100644 --- a/paddle/gserver/layers/AgentLayer.h +++ b/paddle/gserver/layers/AgentLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AverageLayer.cpp b/paddle/gserver/layers/AverageLayer.cpp index af64e15fe3..b8955ab04f 100644 --- a/paddle/gserver/layers/AverageLayer.cpp +++ b/paddle/gserver/layers/AverageLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/AverageLayer.h b/paddle/gserver/layers/AverageLayer.h index 1edc2ace49..b3c4ecec8b 100644 --- a/paddle/gserver/layers/AverageLayer.h +++ b/paddle/gserver/layers/AverageLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BatchNormBaseLayer.cpp b/paddle/gserver/layers/BatchNormBaseLayer.cpp index fd534b2ac4..51463f1118 100644 --- a/paddle/gserver/layers/BatchNormBaseLayer.cpp +++ b/paddle/gserver/layers/BatchNormBaseLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BatchNormBaseLayer.h b/paddle/gserver/layers/BatchNormBaseLayer.h index f956646a6d..f5a555a6d0 100644 --- a/paddle/gserver/layers/BatchNormBaseLayer.h +++ b/paddle/gserver/layers/BatchNormBaseLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BatchNormalizationLayer.cpp b/paddle/gserver/layers/BatchNormalizationLayer.cpp index bdc20c9d81..e6a0624636 100644 --- a/paddle/gserver/layers/BatchNormalizationLayer.cpp +++ b/paddle/gserver/layers/BatchNormalizationLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BatchNormalizationLayer.h b/paddle/gserver/layers/BatchNormalizationLayer.h index 36925a5ed2..56be473568 100644 --- a/paddle/gserver/layers/BatchNormalizationLayer.h +++ b/paddle/gserver/layers/BatchNormalizationLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BilinearInterpLayer.cpp b/paddle/gserver/layers/BilinearInterpLayer.cpp index 11028290dc..1976cb0017 100644 --- a/paddle/gserver/layers/BilinearInterpLayer.cpp +++ b/paddle/gserver/layers/BilinearInterpLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BilinearInterpLayer.h b/paddle/gserver/layers/BilinearInterpLayer.h index eba3c054fa..4ff4b0ea79 100644 --- a/paddle/gserver/layers/BilinearInterpLayer.h +++ b/paddle/gserver/layers/BilinearInterpLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BlockExpandLayer.cpp b/paddle/gserver/layers/BlockExpandLayer.cpp index 17d77879b2..2bafeb9215 100644 --- a/paddle/gserver/layers/BlockExpandLayer.cpp +++ b/paddle/gserver/layers/BlockExpandLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/BlockExpandLayer.h b/paddle/gserver/layers/BlockExpandLayer.h index 1496fb681a..cc96fdd03f 100644 --- a/paddle/gserver/layers/BlockExpandLayer.h +++ b/paddle/gserver/layers/BlockExpandLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CRFDecodingLayer.cpp b/paddle/gserver/layers/CRFDecodingLayer.cpp index 8986741dc3..fdb46aba68 100644 --- a/paddle/gserver/layers/CRFDecodingLayer.cpp +++ b/paddle/gserver/layers/CRFDecodingLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CRFDecodingLayer.h b/paddle/gserver/layers/CRFDecodingLayer.h index 1914062011..1fd444ad10 100644 --- a/paddle/gserver/layers/CRFDecodingLayer.h +++ b/paddle/gserver/layers/CRFDecodingLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CRFLayer.cpp b/paddle/gserver/layers/CRFLayer.cpp index ed4f864ba9..02b7aaf17e 100644 --- a/paddle/gserver/layers/CRFLayer.cpp +++ b/paddle/gserver/layers/CRFLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CRFLayer.h b/paddle/gserver/layers/CRFLayer.h index 21c7fc61e1..d21b32b68c 100644 --- a/paddle/gserver/layers/CRFLayer.h +++ b/paddle/gserver/layers/CRFLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CTCLayer.cpp b/paddle/gserver/layers/CTCLayer.cpp index be5d2c8c75..14ec851551 100644 --- a/paddle/gserver/layers/CTCLayer.cpp +++ b/paddle/gserver/layers/CTCLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CTCLayer.h b/paddle/gserver/layers/CTCLayer.h index 18ba12583b..70d429bad6 100644 --- a/paddle/gserver/layers/CTCLayer.h +++ b/paddle/gserver/layers/CTCLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConcatenateLayer.cpp b/paddle/gserver/layers/ConcatenateLayer.cpp index 910eec8bbc..f6b3d86b8c 100644 --- a/paddle/gserver/layers/ConcatenateLayer.cpp +++ b/paddle/gserver/layers/ConcatenateLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ContextProjection.cpp b/paddle/gserver/layers/ContextProjection.cpp index 30dbf168fb..6080aa51b9 100644 --- a/paddle/gserver/layers/ContextProjection.cpp +++ b/paddle/gserver/layers/ContextProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ContextProjection.h b/paddle/gserver/layers/ContextProjection.h index 188dec0fb3..2df43bd04f 100644 --- a/paddle/gserver/layers/ContextProjection.h +++ b/paddle/gserver/layers/ContextProjection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvBaseLayer.cpp b/paddle/gserver/layers/ConvBaseLayer.cpp index b5a2f8b8e1..473ca24a94 100644 --- a/paddle/gserver/layers/ConvBaseLayer.cpp +++ b/paddle/gserver/layers/ConvBaseLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvBaseLayer.h b/paddle/gserver/layers/ConvBaseLayer.h index 85f57dbe0b..aedf4100e3 100644 --- a/paddle/gserver/layers/ConvBaseLayer.h +++ b/paddle/gserver/layers/ConvBaseLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvOperator.cpp b/paddle/gserver/layers/ConvOperator.cpp index dc06c89dab..3ede98ba4b 100644 --- a/paddle/gserver/layers/ConvOperator.cpp +++ b/paddle/gserver/layers/ConvOperator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvProjection.cpp b/paddle/gserver/layers/ConvProjection.cpp index 5a68fb08da..e72dc37ec8 100644 --- a/paddle/gserver/layers/ConvProjection.cpp +++ b/paddle/gserver/layers/ConvProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvProjection.h b/paddle/gserver/layers/ConvProjection.h index 779fe1455a..c32e5e1d3a 100644 --- a/paddle/gserver/layers/ConvProjection.h +++ b/paddle/gserver/layers/ConvProjection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvShiftLayer.cpp b/paddle/gserver/layers/ConvShiftLayer.cpp index 6e77c1f14e..527d885d86 100644 --- a/paddle/gserver/layers/ConvShiftLayer.cpp +++ b/paddle/gserver/layers/ConvShiftLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ConvexCombinationLayer.cpp b/paddle/gserver/layers/ConvexCombinationLayer.cpp index 7e1fef8bc6..57ff95fe37 100644 --- a/paddle/gserver/layers/ConvexCombinationLayer.cpp +++ b/paddle/gserver/layers/ConvexCombinationLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CosSimLayer.cpp b/paddle/gserver/layers/CosSimLayer.cpp index 894cb5b0d8..254120443d 100644 --- a/paddle/gserver/layers/CosSimLayer.cpp +++ b/paddle/gserver/layers/CosSimLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CosSimLayer.h b/paddle/gserver/layers/CosSimLayer.h index bc47998c11..5dcc5d8a5b 100644 --- a/paddle/gserver/layers/CosSimLayer.h +++ b/paddle/gserver/layers/CosSimLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CosSimVecMatLayer.cpp b/paddle/gserver/layers/CosSimVecMatLayer.cpp index 56d177da64..e8a7f671ee 100644 --- a/paddle/gserver/layers/CosSimVecMatLayer.cpp +++ b/paddle/gserver/layers/CosSimVecMatLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CostLayer.cpp b/paddle/gserver/layers/CostLayer.cpp index 5c839f2d6c..90cd473c42 100644 --- a/paddle/gserver/layers/CostLayer.cpp +++ b/paddle/gserver/layers/CostLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CostLayer.h b/paddle/gserver/layers/CostLayer.h index 120ff9bd2d..7f73bdb3f7 100644 --- a/paddle/gserver/layers/CostLayer.h +++ b/paddle/gserver/layers/CostLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnBatchNormLayer.cpp b/paddle/gserver/layers/CudnnBatchNormLayer.cpp index 6be62b1a25..d44c217105 100644 --- a/paddle/gserver/layers/CudnnBatchNormLayer.cpp +++ b/paddle/gserver/layers/CudnnBatchNormLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnBatchNormLayer.h b/paddle/gserver/layers/CudnnBatchNormLayer.h index 6220e77ceb..a52a683e15 100644 --- a/paddle/gserver/layers/CudnnBatchNormLayer.h +++ b/paddle/gserver/layers/CudnnBatchNormLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnConvLayer.cpp b/paddle/gserver/layers/CudnnConvLayer.cpp index 93c5565d2f..6e28d5eb42 100644 --- a/paddle/gserver/layers/CudnnConvLayer.cpp +++ b/paddle/gserver/layers/CudnnConvLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnConvLayer.h b/paddle/gserver/layers/CudnnConvLayer.h index 6cfbadfb53..6317fab6f8 100644 --- a/paddle/gserver/layers/CudnnConvLayer.h +++ b/paddle/gserver/layers/CudnnConvLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnPoolLayer.cpp b/paddle/gserver/layers/CudnnPoolLayer.cpp index 21d8e2579f..d0e71c6345 100644 --- a/paddle/gserver/layers/CudnnPoolLayer.cpp +++ b/paddle/gserver/layers/CudnnPoolLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/CudnnPoolLayer.h b/paddle/gserver/layers/CudnnPoolLayer.h index 6a6b28db96..072b2f9513 100644 --- a/paddle/gserver/layers/CudnnPoolLayer.h +++ b/paddle/gserver/layers/CudnnPoolLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DataLayer.cpp b/paddle/gserver/layers/DataLayer.cpp index 67c4923036..66f0606a38 100644 --- a/paddle/gserver/layers/DataLayer.cpp +++ b/paddle/gserver/layers/DataLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DataLayer.h b/paddle/gserver/layers/DataLayer.h index da74702201..d3bc97bb6c 100644 --- a/paddle/gserver/layers/DataLayer.h +++ b/paddle/gserver/layers/DataLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DataNormLayer.cpp b/paddle/gserver/layers/DataNormLayer.cpp index b398f3dbed..afd532c949 100644 --- a/paddle/gserver/layers/DataNormLayer.cpp +++ b/paddle/gserver/layers/DataNormLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DataNormLayer.h b/paddle/gserver/layers/DataNormLayer.h index 1179d94fbb..b3043cffd2 100644 --- a/paddle/gserver/layers/DataNormLayer.h +++ b/paddle/gserver/layers/DataNormLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DotMulOperator.cpp b/paddle/gserver/layers/DotMulOperator.cpp index 9409493fda..55dabd79d0 100644 --- a/paddle/gserver/layers/DotMulOperator.cpp +++ b/paddle/gserver/layers/DotMulOperator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/DotMulProjection.cpp b/paddle/gserver/layers/DotMulProjection.cpp index 862eeb6f01..0a1ede3618 100644 --- a/paddle/gserver/layers/DotMulProjection.cpp +++ b/paddle/gserver/layers/DotMulProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/EosIdCheckLayer.cpp b/paddle/gserver/layers/EosIdCheckLayer.cpp index 3a43705d26..dc3c6e6b64 100644 --- a/paddle/gserver/layers/EosIdCheckLayer.cpp +++ b/paddle/gserver/layers/EosIdCheckLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvBaseLayer.cpp b/paddle/gserver/layers/ExpandConvBaseLayer.cpp index 3724609720..25948747fe 100644 --- a/paddle/gserver/layers/ExpandConvBaseLayer.cpp +++ b/paddle/gserver/layers/ExpandConvBaseLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvBaseLayer.h b/paddle/gserver/layers/ExpandConvBaseLayer.h index 5939d27e2a..e14f6e6f44 100644 --- a/paddle/gserver/layers/ExpandConvBaseLayer.h +++ b/paddle/gserver/layers/ExpandConvBaseLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvLayer.cpp b/paddle/gserver/layers/ExpandConvLayer.cpp index 0649289c1c..dcc7839960 100644 --- a/paddle/gserver/layers/ExpandConvLayer.cpp +++ b/paddle/gserver/layers/ExpandConvLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvLayer.h b/paddle/gserver/layers/ExpandConvLayer.h index 82a9e88a42..6f8504b50a 100644 --- a/paddle/gserver/layers/ExpandConvLayer.h +++ b/paddle/gserver/layers/ExpandConvLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvTransLayer.cpp b/paddle/gserver/layers/ExpandConvTransLayer.cpp index 1132ab4f92..cd4965c3c5 100644 --- a/paddle/gserver/layers/ExpandConvTransLayer.cpp +++ b/paddle/gserver/layers/ExpandConvTransLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandConvTransLayer.h b/paddle/gserver/layers/ExpandConvTransLayer.h index 47efe3f656..fa9d7fb481 100644 --- a/paddle/gserver/layers/ExpandConvTransLayer.h +++ b/paddle/gserver/layers/ExpandConvTransLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandLayer.cpp b/paddle/gserver/layers/ExpandLayer.cpp index 9290ce4f6d..de5acfde05 100644 --- a/paddle/gserver/layers/ExpandLayer.cpp +++ b/paddle/gserver/layers/ExpandLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ExpandLayer.h b/paddle/gserver/layers/ExpandLayer.h index fbe0ced9b1..5c63614423 100644 --- a/paddle/gserver/layers/ExpandLayer.h +++ b/paddle/gserver/layers/ExpandLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/FeatureMapExpandLayer.cpp b/paddle/gserver/layers/FeatureMapExpandLayer.cpp index 97c8d143fe..d023074c52 100644 --- a/paddle/gserver/layers/FeatureMapExpandLayer.cpp +++ b/paddle/gserver/layers/FeatureMapExpandLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/FullMatrixProjection.cpp b/paddle/gserver/layers/FullMatrixProjection.cpp index 35a5cb5b7a..9e72a33a3c 100644 --- a/paddle/gserver/layers/FullMatrixProjection.cpp +++ b/paddle/gserver/layers/FullMatrixProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/FullMatrixProjection.h b/paddle/gserver/layers/FullMatrixProjection.h index ddb1e7b18c..58499f2e1e 100644 --- a/paddle/gserver/layers/FullMatrixProjection.h +++ b/paddle/gserver/layers/FullMatrixProjection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/FullyConnectedLayer.cpp b/paddle/gserver/layers/FullyConnectedLayer.cpp index 70c56499a7..d2a028dd80 100644 --- a/paddle/gserver/layers/FullyConnectedLayer.cpp +++ b/paddle/gserver/layers/FullyConnectedLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/FullyConnectedLayer.h b/paddle/gserver/layers/FullyConnectedLayer.h index e15e1236cd..ccd584585c 100644 --- a/paddle/gserver/layers/FullyConnectedLayer.h +++ b/paddle/gserver/layers/FullyConnectedLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GatedRecurrentLayer.cpp b/paddle/gserver/layers/GatedRecurrentLayer.cpp index 495c2174f3..01b210ba70 100644 --- a/paddle/gserver/layers/GatedRecurrentLayer.cpp +++ b/paddle/gserver/layers/GatedRecurrentLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GatedRecurrentLayer.h b/paddle/gserver/layers/GatedRecurrentLayer.h index 3b8706a44e..e099b4d18b 100644 --- a/paddle/gserver/layers/GatedRecurrentLayer.h +++ b/paddle/gserver/layers/GatedRecurrentLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GetOutputLayer.cpp b/paddle/gserver/layers/GetOutputLayer.cpp index 01579d55fd..b77fdbb30e 100644 --- a/paddle/gserver/layers/GetOutputLayer.cpp +++ b/paddle/gserver/layers/GetOutputLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GruCompute.cpp b/paddle/gserver/layers/GruCompute.cpp index d9d423af44..7d4e8001a8 100644 --- a/paddle/gserver/layers/GruCompute.cpp +++ b/paddle/gserver/layers/GruCompute.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GruCompute.cu b/paddle/gserver/layers/GruCompute.cu index 4a3cf6b1ca..d5e547dce3 100644 --- a/paddle/gserver/layers/GruCompute.cu +++ b/paddle/gserver/layers/GruCompute.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GruCompute.h b/paddle/gserver/layers/GruCompute.h index 58b5aacba0..2a5da72068 100644 --- a/paddle/gserver/layers/GruCompute.h +++ b/paddle/gserver/layers/GruCompute.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/GruStepLayer.cpp b/paddle/gserver/layers/GruStepLayer.cpp index 6c9b0c5771..c48b5e40e6 100644 --- a/paddle/gserver/layers/GruStepLayer.cpp +++ b/paddle/gserver/layers/GruStepLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp b/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp index 61bc777785..d62a8d846e 100644 --- a/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp +++ b/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/HierarchicalSigmoidLayer.h b/paddle/gserver/layers/HierarchicalSigmoidLayer.h index 10762bc926..70da3ac126 100644 --- a/paddle/gserver/layers/HierarchicalSigmoidLayer.h +++ b/paddle/gserver/layers/HierarchicalSigmoidLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/IdentityProjection.cpp b/paddle/gserver/layers/IdentityProjection.cpp index b38656c960..8660631b5a 100644 --- a/paddle/gserver/layers/IdentityProjection.cpp +++ b/paddle/gserver/layers/IdentityProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/InterpolationLayer.cpp b/paddle/gserver/layers/InterpolationLayer.cpp index b00bee2356..94d4614b21 100644 --- a/paddle/gserver/layers/InterpolationLayer.cpp +++ b/paddle/gserver/layers/InterpolationLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Layer.cpp b/paddle/gserver/layers/Layer.cpp index a83b0e9ab4..3c539f3076 100644 --- a/paddle/gserver/layers/Layer.cpp +++ b/paddle/gserver/layers/Layer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Layer.h b/paddle/gserver/layers/Layer.h index 3d427a1ac6..6609e16c4c 100644 --- a/paddle/gserver/layers/Layer.h +++ b/paddle/gserver/layers/Layer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LinearChainCRF.cpp b/paddle/gserver/layers/LinearChainCRF.cpp index e2a4f69e71..c6414c822e 100644 --- a/paddle/gserver/layers/LinearChainCRF.cpp +++ b/paddle/gserver/layers/LinearChainCRF.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LinearChainCRF.h b/paddle/gserver/layers/LinearChainCRF.h index 6368f2b9de..a905bf803d 100644 --- a/paddle/gserver/layers/LinearChainCRF.h +++ b/paddle/gserver/layers/LinearChainCRF.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LinearChainCTC.cpp b/paddle/gserver/layers/LinearChainCTC.cpp index 3368eb4d8a..60e814fc30 100644 --- a/paddle/gserver/layers/LinearChainCTC.cpp +++ b/paddle/gserver/layers/LinearChainCTC.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LinearChainCTC.h b/paddle/gserver/layers/LinearChainCTC.h index 0a93d2e9a6..737c9d5c31 100644 --- a/paddle/gserver/layers/LinearChainCTC.h +++ b/paddle/gserver/layers/LinearChainCTC.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmCompute.cpp b/paddle/gserver/layers/LstmCompute.cpp index 38057636ed..18f7996958 100644 --- a/paddle/gserver/layers/LstmCompute.cpp +++ b/paddle/gserver/layers/LstmCompute.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmCompute.cu b/paddle/gserver/layers/LstmCompute.cu index af271d682f..f75c0c40cc 100644 --- a/paddle/gserver/layers/LstmCompute.cu +++ b/paddle/gserver/layers/LstmCompute.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmCompute.h b/paddle/gserver/layers/LstmCompute.h index 97be7218f2..9b7aee19dd 100644 --- a/paddle/gserver/layers/LstmCompute.h +++ b/paddle/gserver/layers/LstmCompute.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmLayer.cpp b/paddle/gserver/layers/LstmLayer.cpp index e70a20e5c0..975edcfe7f 100644 --- a/paddle/gserver/layers/LstmLayer.cpp +++ b/paddle/gserver/layers/LstmLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmLayer.h b/paddle/gserver/layers/LstmLayer.h index 5b936ff44e..16c62aa88d 100644 --- a/paddle/gserver/layers/LstmLayer.h +++ b/paddle/gserver/layers/LstmLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/LstmStepLayer.cpp b/paddle/gserver/layers/LstmStepLayer.cpp index e7a8d519f2..5fc6474b86 100644 --- a/paddle/gserver/layers/LstmStepLayer.cpp +++ b/paddle/gserver/layers/LstmStepLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MDLstmLayer.cpp b/paddle/gserver/layers/MDLstmLayer.cpp index 93f52c1c31..9d3797d16f 100644 --- a/paddle/gserver/layers/MDLstmLayer.cpp +++ b/paddle/gserver/layers/MDLstmLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MaxIdLayer.cpp b/paddle/gserver/layers/MaxIdLayer.cpp index 22670fa121..80555f3f7b 100644 --- a/paddle/gserver/layers/MaxIdLayer.cpp +++ b/paddle/gserver/layers/MaxIdLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MaxLayer.cpp b/paddle/gserver/layers/MaxLayer.cpp index 42bc6bb815..23629e1986 100644 --- a/paddle/gserver/layers/MaxLayer.cpp +++ b/paddle/gserver/layers/MaxLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MaxLayer.h b/paddle/gserver/layers/MaxLayer.h index 74df0b8b57..472ee0ccca 100644 --- a/paddle/gserver/layers/MaxLayer.h +++ b/paddle/gserver/layers/MaxLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MaxOutLayer.cpp b/paddle/gserver/layers/MaxOutLayer.cpp index b7f1b98041..4fb99ce2a2 100644 --- a/paddle/gserver/layers/MaxOutLayer.cpp +++ b/paddle/gserver/layers/MaxOutLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MaxOutLayer.h b/paddle/gserver/layers/MaxOutLayer.h index 9011a5c332..59c2245e0d 100644 --- a/paddle/gserver/layers/MaxOutLayer.h +++ b/paddle/gserver/layers/MaxOutLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MixedLayer.cpp b/paddle/gserver/layers/MixedLayer.cpp index 1392188fca..490b217347 100644 --- a/paddle/gserver/layers/MixedLayer.cpp +++ b/paddle/gserver/layers/MixedLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MixedLayer.h b/paddle/gserver/layers/MixedLayer.h index 271e0c2538..d73ba6b7a1 100644 --- a/paddle/gserver/layers/MixedLayer.h +++ b/paddle/gserver/layers/MixedLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MultinomialSampler.cpp b/paddle/gserver/layers/MultinomialSampler.cpp index e85dca72d3..0b285ed20f 100644 --- a/paddle/gserver/layers/MultinomialSampler.cpp +++ b/paddle/gserver/layers/MultinomialSampler.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MultinomialSampler.h b/paddle/gserver/layers/MultinomialSampler.h index 59683d2ee2..6e50f8738e 100644 --- a/paddle/gserver/layers/MultinomialSampler.h +++ b/paddle/gserver/layers/MultinomialSampler.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/MultiplexLayer.cpp b/paddle/gserver/layers/MultiplexLayer.cpp index c681eb0623..dc4a1ec321 100644 --- a/paddle/gserver/layers/MultiplexLayer.cpp +++ b/paddle/gserver/layers/MultiplexLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/NCELayer.cpp b/paddle/gserver/layers/NCELayer.cpp index 50b29cdea5..540db46545 100644 --- a/paddle/gserver/layers/NCELayer.cpp +++ b/paddle/gserver/layers/NCELayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/NormLayer.cpp b/paddle/gserver/layers/NormLayer.cpp index 445a1a0c52..b8682a1422 100644 --- a/paddle/gserver/layers/NormLayer.cpp +++ b/paddle/gserver/layers/NormLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/NormLayer.h b/paddle/gserver/layers/NormLayer.h index fcc57849d6..aedbb95b4f 100644 --- a/paddle/gserver/layers/NormLayer.h +++ b/paddle/gserver/layers/NormLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/NormProjectionLayer.cpp b/paddle/gserver/layers/NormProjectionLayer.cpp index da36cc2c99..ea301292e0 100644 --- a/paddle/gserver/layers/NormProjectionLayer.cpp +++ b/paddle/gserver/layers/NormProjectionLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/NormProjectionLayer.h b/paddle/gserver/layers/NormProjectionLayer.h index b42e98ab09..0db8e2551f 100644 --- a/paddle/gserver/layers/NormProjectionLayer.h +++ b/paddle/gserver/layers/NormProjectionLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Operator.cpp b/paddle/gserver/layers/Operator.cpp index b89c474014..a638933914 100644 --- a/paddle/gserver/layers/Operator.cpp +++ b/paddle/gserver/layers/Operator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Operator.h b/paddle/gserver/layers/Operator.h index ff6558dc73..b0586b59e9 100644 --- a/paddle/gserver/layers/Operator.h +++ b/paddle/gserver/layers/Operator.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/OuterProdLayer.cpp b/paddle/gserver/layers/OuterProdLayer.cpp index 9b24a4f440..42587dcce5 100644 --- a/paddle/gserver/layers/OuterProdLayer.cpp +++ b/paddle/gserver/layers/OuterProdLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ParameterReluLayer.cpp b/paddle/gserver/layers/ParameterReluLayer.cpp index cd3bffa2e1..836c1981ba 100644 --- a/paddle/gserver/layers/ParameterReluLayer.cpp +++ b/paddle/gserver/layers/ParameterReluLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ParameterReluLayer.h b/paddle/gserver/layers/ParameterReluLayer.h index 029c09381f..a82497fc01 100644 --- a/paddle/gserver/layers/ParameterReluLayer.h +++ b/paddle/gserver/layers/ParameterReluLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolLayer.cpp b/paddle/gserver/layers/PoolLayer.cpp index 511dfd87c1..36e396487e 100644 --- a/paddle/gserver/layers/PoolLayer.cpp +++ b/paddle/gserver/layers/PoolLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolLayer.h b/paddle/gserver/layers/PoolLayer.h index 59be295a53..c05d7a364d 100644 --- a/paddle/gserver/layers/PoolLayer.h +++ b/paddle/gserver/layers/PoolLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolProjection.cpp b/paddle/gserver/layers/PoolProjection.cpp index 1b227c8084..d90b438448 100644 --- a/paddle/gserver/layers/PoolProjection.cpp +++ b/paddle/gserver/layers/PoolProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolProjection.h b/paddle/gserver/layers/PoolProjection.h index 9c3191bd80..9a75f465f6 100644 --- a/paddle/gserver/layers/PoolProjection.h +++ b/paddle/gserver/layers/PoolProjection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolProjectionLayer.cpp b/paddle/gserver/layers/PoolProjectionLayer.cpp index aabc60af19..392c548d45 100644 --- a/paddle/gserver/layers/PoolProjectionLayer.cpp +++ b/paddle/gserver/layers/PoolProjectionLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PoolProjectionLayer.h b/paddle/gserver/layers/PoolProjectionLayer.h index 777b6f39e7..3dc6af2f0e 100644 --- a/paddle/gserver/layers/PoolProjectionLayer.h +++ b/paddle/gserver/layers/PoolProjectionLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PowerLayer.cpp b/paddle/gserver/layers/PowerLayer.cpp index dbe70a1d87..eb69249270 100644 --- a/paddle/gserver/layers/PowerLayer.cpp +++ b/paddle/gserver/layers/PowerLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/PrintLayer.cpp b/paddle/gserver/layers/PrintLayer.cpp index 95be7b34cb..ac7f658864 100644 --- a/paddle/gserver/layers/PrintLayer.cpp +++ b/paddle/gserver/layers/PrintLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Projection.cpp b/paddle/gserver/layers/Projection.cpp index c7eb4b6442..974b3cf059 100644 --- a/paddle/gserver/layers/Projection.cpp +++ b/paddle/gserver/layers/Projection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/Projection.h b/paddle/gserver/layers/Projection.h index 798503113d..8cd8042479 100644 --- a/paddle/gserver/layers/Projection.h +++ b/paddle/gserver/layers/Projection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/RecurrentLayer.cpp b/paddle/gserver/layers/RecurrentLayer.cpp index 08453e21b8..0832eeaa10 100644 --- a/paddle/gserver/layers/RecurrentLayer.cpp +++ b/paddle/gserver/layers/RecurrentLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/RecurrentLayerGroup.cpp b/paddle/gserver/layers/RecurrentLayerGroup.cpp index a5443975da..5cb4220623 100644 --- a/paddle/gserver/layers/RecurrentLayerGroup.cpp +++ b/paddle/gserver/layers/RecurrentLayerGroup.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ResizeLayer.cpp b/paddle/gserver/layers/ResizeLayer.cpp index 3c478a33e3..e79732155a 100644 --- a/paddle/gserver/layers/ResizeLayer.cpp +++ b/paddle/gserver/layers/ResizeLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SamplingIdLayer.cpp b/paddle/gserver/layers/SamplingIdLayer.cpp index b39c9948b5..59ff5d41b5 100644 --- a/paddle/gserver/layers/SamplingIdLayer.cpp +++ b/paddle/gserver/layers/SamplingIdLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ScalingLayer.cpp b/paddle/gserver/layers/ScalingLayer.cpp index 71570810f9..013bff6b98 100644 --- a/paddle/gserver/layers/ScalingLayer.cpp +++ b/paddle/gserver/layers/ScalingLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ScalingProjection.cpp b/paddle/gserver/layers/ScalingProjection.cpp index 7999d02d38..ddb8c87110 100644 --- a/paddle/gserver/layers/ScalingProjection.cpp +++ b/paddle/gserver/layers/ScalingProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp b/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp index 4dfa2c179d..75d9fa8a97 100644 --- a/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp +++ b/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SelectiveFullyConnectedLayer.h b/paddle/gserver/layers/SelectiveFullyConnectedLayer.h index 9f92ae0605..bdf9a4652c 100644 --- a/paddle/gserver/layers/SelectiveFullyConnectedLayer.h +++ b/paddle/gserver/layers/SelectiveFullyConnectedLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequenceConcatLayer.cpp b/paddle/gserver/layers/SequenceConcatLayer.cpp index bd72ba3d16..d3e0e16e96 100644 --- a/paddle/gserver/layers/SequenceConcatLayer.cpp +++ b/paddle/gserver/layers/SequenceConcatLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequenceLastInstanceLayer.cpp b/paddle/gserver/layers/SequenceLastInstanceLayer.cpp index 0e9531eabb..4bfce766c7 100644 --- a/paddle/gserver/layers/SequenceLastInstanceLayer.cpp +++ b/paddle/gserver/layers/SequenceLastInstanceLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequencePoolLayer.cpp b/paddle/gserver/layers/SequencePoolLayer.cpp index c9f19b7d3b..856c889e3b 100644 --- a/paddle/gserver/layers/SequencePoolLayer.cpp +++ b/paddle/gserver/layers/SequencePoolLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequencePoolLayer.h b/paddle/gserver/layers/SequencePoolLayer.h index 669af80e1d..aa9c132586 100644 --- a/paddle/gserver/layers/SequencePoolLayer.h +++ b/paddle/gserver/layers/SequencePoolLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequenceReshapeLayer.cpp b/paddle/gserver/layers/SequenceReshapeLayer.cpp index 5ca9b8b300..4b90424215 100644 --- a/paddle/gserver/layers/SequenceReshapeLayer.cpp +++ b/paddle/gserver/layers/SequenceReshapeLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequenceToBatch.cpp b/paddle/gserver/layers/SequenceToBatch.cpp index 04402db9c8..c12ed82197 100644 --- a/paddle/gserver/layers/SequenceToBatch.cpp +++ b/paddle/gserver/layers/SequenceToBatch.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SequenceToBatch.h b/paddle/gserver/layers/SequenceToBatch.h index 6bc12f207e..fe9b34b224 100644 --- a/paddle/gserver/layers/SequenceToBatch.h +++ b/paddle/gserver/layers/SequenceToBatch.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SlopeInterceptLayer.cpp b/paddle/gserver/layers/SlopeInterceptLayer.cpp index dd6ffcd50b..5c00e54f8c 100644 --- a/paddle/gserver/layers/SlopeInterceptLayer.cpp +++ b/paddle/gserver/layers/SlopeInterceptLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp b/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp index dce660a5bc..14fe88ff8a 100644 --- a/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp +++ b/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SpatialPyramidPoolLayer.h b/paddle/gserver/layers/SpatialPyramidPoolLayer.h index 79db574d99..32e88cf141 100644 --- a/paddle/gserver/layers/SpatialPyramidPoolLayer.h +++ b/paddle/gserver/layers/SpatialPyramidPoolLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SubSequenceLayer.cpp b/paddle/gserver/layers/SubSequenceLayer.cpp index 664f9e13c0..8b35456391 100644 --- a/paddle/gserver/layers/SubSequenceLayer.cpp +++ b/paddle/gserver/layers/SubSequenceLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/SumToOneNormLayer.cpp b/paddle/gserver/layers/SumToOneNormLayer.cpp index bcf3916840..e6759171cb 100644 --- a/paddle/gserver/layers/SumToOneNormLayer.cpp +++ b/paddle/gserver/layers/SumToOneNormLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TableProjection.cpp b/paddle/gserver/layers/TableProjection.cpp index 2bc0d329d9..270acdd34b 100644 --- a/paddle/gserver/layers/TableProjection.cpp +++ b/paddle/gserver/layers/TableProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TableProjection.h b/paddle/gserver/layers/TableProjection.h index 97c672508a..fb6c0e17c2 100644 --- a/paddle/gserver/layers/TableProjection.h +++ b/paddle/gserver/layers/TableProjection.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TensorLayer.cpp b/paddle/gserver/layers/TensorLayer.cpp index 03586cc6ff..642eb1bdd3 100644 --- a/paddle/gserver/layers/TensorLayer.cpp +++ b/paddle/gserver/layers/TensorLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TensorLayer.h b/paddle/gserver/layers/TensorLayer.h index 9ac651de4d..ac38ffb620 100644 --- a/paddle/gserver/layers/TensorLayer.h +++ b/paddle/gserver/layers/TensorLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TransLayer.cpp b/paddle/gserver/layers/TransLayer.cpp index 53a24d4cc4..5cbaaf8f08 100644 --- a/paddle/gserver/layers/TransLayer.cpp +++ b/paddle/gserver/layers/TransLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TransLayer.h b/paddle/gserver/layers/TransLayer.h index 25b091f9f4..8189700759 100644 --- a/paddle/gserver/layers/TransLayer.h +++ b/paddle/gserver/layers/TransLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/TransposedFullMatrixProjection.cpp b/paddle/gserver/layers/TransposedFullMatrixProjection.cpp index c883283f78..8282584ab4 100644 --- a/paddle/gserver/layers/TransposedFullMatrixProjection.cpp +++ b/paddle/gserver/layers/TransposedFullMatrixProjection.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ValidationLayer.cpp b/paddle/gserver/layers/ValidationLayer.cpp index 0fee4bd246..f029ea4c51 100644 --- a/paddle/gserver/layers/ValidationLayer.cpp +++ b/paddle/gserver/layers/ValidationLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/ValidationLayer.h b/paddle/gserver/layers/ValidationLayer.h index eef9c80a7b..f9c61503aa 100644 --- a/paddle/gserver/layers/ValidationLayer.h +++ b/paddle/gserver/layers/ValidationLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/WarpCTCLayer.cpp b/paddle/gserver/layers/WarpCTCLayer.cpp index e68363a1b2..23ca5257b6 100644 --- a/paddle/gserver/layers/WarpCTCLayer.cpp +++ b/paddle/gserver/layers/WarpCTCLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/layers/WarpCTCLayer.h b/paddle/gserver/layers/WarpCTCLayer.h index 1b0f5ba267..3d9ae9249a 100644 --- a/paddle/gserver/layers/WarpCTCLayer.h +++ b/paddle/gserver/layers/WarpCTCLayer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/LayerGradUtil.cpp b/paddle/gserver/tests/LayerGradUtil.cpp index 4757516917..dffc24936f 100644 --- a/paddle/gserver/tests/LayerGradUtil.cpp +++ b/paddle/gserver/tests/LayerGradUtil.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/LayerGradUtil.h b/paddle/gserver/tests/LayerGradUtil.h index a061c7fc53..2b8f334f19 100644 --- a/paddle/gserver/tests/LayerGradUtil.h +++ b/paddle/gserver/tests/LayerGradUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/TestUtil.cpp b/paddle/gserver/tests/TestUtil.cpp index 84d516683c..dc00711697 100644 --- a/paddle/gserver/tests/TestUtil.cpp +++ b/paddle/gserver/tests/TestUtil.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/TestUtil.h b/paddle/gserver/tests/TestUtil.h index 000f8884e8..ec86469aeb 100644 --- a/paddle/gserver/tests/TestUtil.h +++ b/paddle/gserver/tests/TestUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/__init__.py b/paddle/gserver/tests/__init__.py index c90af2ee00..f662d68263 100644 --- a/paddle/gserver/tests/__init__.py +++ b/paddle/gserver/tests/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_dotmul_a.conf b/paddle/gserver/tests/concat_dotmul_a.conf index 52340596b9..db02ca7e80 100644 --- a/paddle/gserver/tests/concat_dotmul_a.conf +++ b/paddle/gserver/tests/concat_dotmul_a.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_dotmul_b.conf b/paddle/gserver/tests/concat_dotmul_b.conf index 68859867bf..5e64970e44 100644 --- a/paddle/gserver/tests/concat_dotmul_b.conf +++ b/paddle/gserver/tests/concat_dotmul_b.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_fullmatrix_a.conf b/paddle/gserver/tests/concat_fullmatrix_a.conf index 35bafc58ac..940d1efc58 100644 --- a/paddle/gserver/tests/concat_fullmatrix_a.conf +++ b/paddle/gserver/tests/concat_fullmatrix_a.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_fullmatrix_b.conf b/paddle/gserver/tests/concat_fullmatrix_b.conf index 00a957d97d..931e5b38ef 100644 --- a/paddle/gserver/tests/concat_fullmatrix_b.conf +++ b/paddle/gserver/tests/concat_fullmatrix_b.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_table_a.conf b/paddle/gserver/tests/concat_table_a.conf index a8ff70f883..047cb44d15 100644 --- a/paddle/gserver/tests/concat_table_a.conf +++ b/paddle/gserver/tests/concat_table_a.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/concat_table_b.conf b/paddle/gserver/tests/concat_table_b.conf index 95d7c10f7b..c666ab9942 100644 --- a/paddle/gserver/tests/concat_table_b.conf +++ b/paddle/gserver/tests/concat_table_b.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/img_conv_a.conf b/paddle/gserver/tests/img_conv_a.conf index 20c89b875e..3ad15c64fe 100644 --- a/paddle/gserver/tests/img_conv_a.conf +++ b/paddle/gserver/tests/img_conv_a.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/img_conv_b.conf b/paddle/gserver/tests/img_conv_b.conf index 19b99c77fd..e68008155e 100644 --- a/paddle/gserver/tests/img_conv_b.conf +++ b/paddle/gserver/tests/img_conv_b.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/img_conv_c.conf b/paddle/gserver/tests/img_conv_c.conf index fea332f6d1..4598ffbdb2 100644 --- a/paddle/gserver/tests/img_conv_c.conf +++ b/paddle/gserver/tests/img_conv_c.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/img_pool_a.conf b/paddle/gserver/tests/img_pool_a.conf index 9bd046b533..afd271055d 100644 --- a/paddle/gserver/tests/img_pool_a.conf +++ b/paddle/gserver/tests/img_pool_a.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/img_pool_b.conf b/paddle/gserver/tests/img_pool_b.conf index 6ea9649b3f..e8deb9edbe 100644 --- a/paddle/gserver/tests/img_pool_b.conf +++ b/paddle/gserver/tests/img_pool_b.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/pyDataProvider.py b/paddle/gserver/tests/pyDataProvider.py index 91863b4175..7235a23943 100644 --- a/paddle/gserver/tests/pyDataProvider.py +++ b/paddle/gserver/tests/pyDataProvider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/pyDataProvider/trainer.conf b/paddle/gserver/tests/pyDataProvider/trainer.conf index 7957814c01..7d910df20d 100644 --- a/paddle/gserver/tests/pyDataProvider/trainer.conf +++ b/paddle/gserver/tests/pyDataProvider/trainer.conf @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/rnn_data_provider.py b/paddle/gserver/tests/rnn_data_provider.py index 715ac08a42..3afd45c72f 100644 --- a/paddle/gserver/tests/rnn_data_provider.py +++ b/paddle/gserver/tests/rnn_data_provider.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequenceGen.py b/paddle/gserver/tests/sequenceGen.py index 99440ada53..fd725727c0 100644 --- a/paddle/gserver/tests/sequenceGen.py +++ b/paddle/gserver/tests/sequenceGen.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_layer_group.conf b/paddle/gserver/tests/sequence_layer_group.conf index 087aa96ccb..68d150d553 100644 --- a/paddle/gserver/tests/sequence_layer_group.conf +++ b/paddle/gserver/tests/sequence_layer_group.conf @@ -1,5 +1,5 @@ #!/usr/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_nest_layer_group.conf b/paddle/gserver/tests/sequence_nest_layer_group.conf index 93a0f6da79..88cb42798b 100644 --- a/paddle/gserver/tests/sequence_nest_layer_group.conf +++ b/paddle/gserver/tests/sequence_nest_layer_group.conf @@ -1,5 +1,5 @@ #!/usr/bin/env python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_nest_rnn.conf b/paddle/gserver/tests/sequence_nest_rnn.conf index 524760be76..2873a59966 100644 --- a/paddle/gserver/tests/sequence_nest_rnn.conf +++ b/paddle/gserver/tests/sequence_nest_rnn.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf b/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf index 0614958b47..ad14a2c927 100644 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py index 163fce956e..7303d08804 100644 --- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py +++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py @@ -1,5 +1,5 @@ # edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_rnn.conf b/paddle/gserver/tests/sequence_rnn.conf index 3294c2c3fc..1084edfe70 100644 --- a/paddle/gserver/tests/sequence_rnn.conf +++ b/paddle/gserver/tests/sequence_rnn.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_rnn_multi_input.conf b/paddle/gserver/tests/sequence_rnn_multi_input.conf index 51881e21d9..40d0317415 100644 --- a/paddle/gserver/tests/sequence_rnn_multi_input.conf +++ b/paddle/gserver/tests/sequence_rnn_multi_input.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py b/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py index 4cf7035477..786a0c6d78 100644 --- a/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py +++ b/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_ActivationGrad.cpp b/paddle/gserver/tests/test_ActivationGrad.cpp index e54c5109e7..0181d62519 100644 --- a/paddle/gserver/tests/test_ActivationGrad.cpp +++ b/paddle/gserver/tests/test_ActivationGrad.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_BatchNorm.cpp b/paddle/gserver/tests/test_BatchNorm.cpp index 0cb6f58dc0..8575999aba 100644 --- a/paddle/gserver/tests/test_BatchNorm.cpp +++ b/paddle/gserver/tests/test_BatchNorm.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_ConvTrans.cpp b/paddle/gserver/tests/test_ConvTrans.cpp index f3efdfb428..3af3f08f40 100644 --- a/paddle/gserver/tests/test_ConvTrans.cpp +++ b/paddle/gserver/tests/test_ConvTrans.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_ConvUnify.cpp b/paddle/gserver/tests/test_ConvUnify.cpp index 5acf02bea0..d59acf96ac 100644 --- a/paddle/gserver/tests/test_ConvUnify.cpp +++ b/paddle/gserver/tests/test_ConvUnify.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_Evaluator.cpp b/paddle/gserver/tests/test_Evaluator.cpp index be639ea093..2c20f3a52f 100644 --- a/paddle/gserver/tests/test_Evaluator.cpp +++ b/paddle/gserver/tests/test_Evaluator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_LayerGrad.cpp b/paddle/gserver/tests/test_LayerGrad.cpp index 099e96aa6c..7983d9fe64 100644 --- a/paddle/gserver/tests/test_LayerGrad.cpp +++ b/paddle/gserver/tests/test_LayerGrad.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_LinearChainCRF.cpp b/paddle/gserver/tests/test_LinearChainCRF.cpp index 913d6ed751..330adee8f7 100644 --- a/paddle/gserver/tests/test_LinearChainCRF.cpp +++ b/paddle/gserver/tests/test_LinearChainCRF.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_MultinomialSampler.cpp b/paddle/gserver/tests/test_MultinomialSampler.cpp index 3fc099adbd..fc164da8ea 100644 --- a/paddle/gserver/tests/test_MultinomialSampler.cpp +++ b/paddle/gserver/tests/test_MultinomialSampler.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_NetworkCompare.cpp b/paddle/gserver/tests/test_NetworkCompare.cpp index 71ed3bc4b6..ff6b5ab0d0 100644 --- a/paddle/gserver/tests/test_NetworkCompare.cpp +++ b/paddle/gserver/tests/test_NetworkCompare.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_ProtoDataProvider.cpp b/paddle/gserver/tests/test_ProtoDataProvider.cpp index 01070bc1cb..d5b8017cd1 100644 --- a/paddle/gserver/tests/test_ProtoDataProvider.cpp +++ b/paddle/gserver/tests/test_ProtoDataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_PyDataProvider.cpp b/paddle/gserver/tests/test_PyDataProvider.cpp index 802f9aa4cb..0f264ecf91 100644 --- a/paddle/gserver/tests/test_PyDataProvider.cpp +++ b/paddle/gserver/tests/test_PyDataProvider.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_PyDataProvider2.cpp b/paddle/gserver/tests/test_PyDataProvider2.cpp index 6674e6b87c..436318d356 100644 --- a/paddle/gserver/tests/test_PyDataProvider2.cpp +++ b/paddle/gserver/tests/test_PyDataProvider2.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_PyDataProvider2.py b/paddle/gserver/tests/test_PyDataProvider2.py index bf23c52fd7..f7b540013e 100644 --- a/paddle/gserver/tests/test_PyDataProvider2.py +++ b/paddle/gserver/tests/test_PyDataProvider2.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_RecurrentGradientMachine.cpp b/paddle/gserver/tests/test_RecurrentGradientMachine.cpp index 9d86067fb5..a351667d8b 100644 --- a/paddle/gserver/tests/test_RecurrentGradientMachine.cpp +++ b/paddle/gserver/tests/test_RecurrentGradientMachine.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_RecurrentLayer.cpp b/paddle/gserver/tests/test_RecurrentLayer.cpp index 0643cec38b..3f26b710e9 100644 --- a/paddle/gserver/tests/test_RecurrentLayer.cpp +++ b/paddle/gserver/tests/test_RecurrentLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_SelectiveFCLayer.cpp b/paddle/gserver/tests/test_SelectiveFCLayer.cpp index 204b03332f..c588f69446 100644 --- a/paddle/gserver/tests/test_SelectiveFCLayer.cpp +++ b/paddle/gserver/tests/test_SelectiveFCLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/gserver/tests/test_WarpCTCLayer.cpp b/paddle/gserver/tests/test_WarpCTCLayer.cpp index 2dd83db345..e526a27906 100644 --- a/paddle/gserver/tests/test_WarpCTCLayer.cpp +++ b/paddle/gserver/tests/test_WarpCTCLayer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Allocator.h b/paddle/math/Allocator.h index cba8b37289..4d0a1506be 100644 --- a/paddle/math/Allocator.h +++ b/paddle/math/Allocator.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/BaseMatrix.cu b/paddle/math/BaseMatrix.cu index 05faeff2e4..0a0d92d1ae 100644 --- a/paddle/math/BaseMatrix.cu +++ b/paddle/math/BaseMatrix.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/BaseMatrix.h b/paddle/math/BaseMatrix.h index f4576985b8..368557bb26 100644 --- a/paddle/math/BaseMatrix.h +++ b/paddle/math/BaseMatrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/CpuSparseMatrix.cpp b/paddle/math/CpuSparseMatrix.cpp index ad3f8e64ef..324c7ec0ca 100644 --- a/paddle/math/CpuSparseMatrix.cpp +++ b/paddle/math/CpuSparseMatrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/CpuSparseMatrix.h b/paddle/math/CpuSparseMatrix.h index 50f3c1569a..9676f8864f 100644 --- a/paddle/math/CpuSparseMatrix.h +++ b/paddle/math/CpuSparseMatrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/ExecViaCpu.h b/paddle/math/ExecViaCpu.h index 67fb6c0cda..1e03cc5f45 100644 --- a/paddle/math/ExecViaCpu.h +++ b/paddle/math/ExecViaCpu.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MathFunctions.cpp b/paddle/math/MathFunctions.cpp index 1217163bee..037525b402 100644 --- a/paddle/math/MathFunctions.cpp +++ b/paddle/math/MathFunctions.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MathFunctions.h b/paddle/math/MathFunctions.h index 0741c45678..c8559eefd8 100644 --- a/paddle/math/MathFunctions.h +++ b/paddle/math/MathFunctions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MathUtils.cpp b/paddle/math/MathUtils.cpp index 878e0b8723..1fb7655c5a 100644 --- a/paddle/math/MathUtils.cpp +++ b/paddle/math/MathUtils.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MathUtils.h b/paddle/math/MathUtils.h index 907116c002..f2b2980138 100644 --- a/paddle/math/MathUtils.h +++ b/paddle/math/MathUtils.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Matrix.cpp b/paddle/math/Matrix.cpp index b70b47a5fc..c69e074a76 100644 --- a/paddle/math/Matrix.cpp +++ b/paddle/math/Matrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Matrix.h b/paddle/math/Matrix.h index 5de78bb84c..395143a4b1 100644 --- a/paddle/math/Matrix.h +++ b/paddle/math/Matrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MatrixBitCode.cpp b/paddle/math/MatrixBitCode.cpp index ac5b10c7bd..6390d4b6a5 100644 --- a/paddle/math/MatrixBitCode.cpp +++ b/paddle/math/MatrixBitCode.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MemoryHandle.cpp b/paddle/math/MemoryHandle.cpp index 9101957fc6..4c4a827b23 100644 --- a/paddle/math/MemoryHandle.cpp +++ b/paddle/math/MemoryHandle.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/MemoryHandle.h b/paddle/math/MemoryHandle.h index f12635d5d4..0828d377c9 100644 --- a/paddle/math/MemoryHandle.h +++ b/paddle/math/MemoryHandle.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/PoolAllocator.cpp b/paddle/math/PoolAllocator.cpp index 2c150949dd..4282c7243a 100644 --- a/paddle/math/PoolAllocator.cpp +++ b/paddle/math/PoolAllocator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/PoolAllocator.h b/paddle/math/PoolAllocator.h index 5d33b45312..1544cb2cfc 100644 --- a/paddle/math/PoolAllocator.h +++ b/paddle/math/PoolAllocator.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SIMDFunctions.cpp b/paddle/math/SIMDFunctions.cpp index 1fb156f29b..95219debf5 100644 --- a/paddle/math/SIMDFunctions.cpp +++ b/paddle/math/SIMDFunctions.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SIMDFunctions.h b/paddle/math/SIMDFunctions.h index ac82f10910..9b0a8719b2 100644 --- a/paddle/math/SIMDFunctions.h +++ b/paddle/math/SIMDFunctions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SparseMatrix.cpp b/paddle/math/SparseMatrix.cpp index 2b0bff9535..d2779cc9f5 100644 --- a/paddle/math/SparseMatrix.cpp +++ b/paddle/math/SparseMatrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SparseMatrix.h b/paddle/math/SparseMatrix.h index 175ef54b85..f8d9ffc29f 100644 --- a/paddle/math/SparseMatrix.h +++ b/paddle/math/SparseMatrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SparseRowMatrix.cpp b/paddle/math/SparseRowMatrix.cpp index 100827e321..3091743123 100644 --- a/paddle/math/SparseRowMatrix.cpp +++ b/paddle/math/SparseRowMatrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/SparseRowMatrix.h b/paddle/math/SparseRowMatrix.h index 56f113a361..2fee1b39fe 100644 --- a/paddle/math/SparseRowMatrix.h +++ b/paddle/math/SparseRowMatrix.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Storage.cpp b/paddle/math/Storage.cpp index 57ea5c9266..0170b4efb8 100644 --- a/paddle/math/Storage.cpp +++ b/paddle/math/Storage.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Storage.h b/paddle/math/Storage.h index 725de247e6..3658320182 100644 --- a/paddle/math/Storage.h +++ b/paddle/math/Storage.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TensorApply.h b/paddle/math/TensorApply.h index 8b2a9a7cd2..11c7acb441 100644 --- a/paddle/math/TensorApply.h +++ b/paddle/math/TensorApply.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TensorAssign.h b/paddle/math/TensorAssign.h index 03f7048d2d..943fb5649e 100644 --- a/paddle/math/TensorAssign.h +++ b/paddle/math/TensorAssign.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TensorEvaluate.h b/paddle/math/TensorEvaluate.h index 39981246f0..346ed7ab13 100644 --- a/paddle/math/TensorEvaluate.h +++ b/paddle/math/TensorEvaluate.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TensorExpression.h b/paddle/math/TensorExpression.h index b28ea2be1d..7f28ad83bb 100644 --- a/paddle/math/TensorExpression.h +++ b/paddle/math/TensorExpression.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TrainingAlgorithmOp.cu b/paddle/math/TrainingAlgorithmOp.cu index d8d9c793fb..72ff077270 100644 --- a/paddle/math/TrainingAlgorithmOp.cu +++ b/paddle/math/TrainingAlgorithmOp.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/TrainingAlgorithmOp.h b/paddle/math/TrainingAlgorithmOp.h index 68eb98a93e..2dc56f69e5 100644 --- a/paddle/math/TrainingAlgorithmOp.h +++ b/paddle/math/TrainingAlgorithmOp.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Vector.cpp b/paddle/math/Vector.cpp index b2ade83138..484f4c9252 100644 --- a/paddle/math/Vector.cpp +++ b/paddle/math/Vector.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/Vector.h b/paddle/math/Vector.h index bcd8ff3fa3..535580ac37 100644 --- a/paddle/math/Vector.h +++ b/paddle/math/Vector.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/OriginalOptimizerApi.h b/paddle/math/tests/OriginalOptimizerApi.h index fe4d1ae542..ddcdd6bb51 100644 --- a/paddle/math/tests/OriginalOptimizerApi.h +++ b/paddle/math/tests/OriginalOptimizerApi.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/PerfUtils.h b/paddle/math/tests/PerfUtils.h index c32f4c634a..9c6a63ce6c 100644 --- a/paddle/math/tests/PerfUtils.h +++ b/paddle/math/tests/PerfUtils.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/TensorCheck.h b/paddle/math/tests/TensorCheck.h index 956bcf61a4..5bc4a03067 100644 --- a/paddle/math/tests/TensorCheck.h +++ b/paddle/math/tests/TensorCheck.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/TestUtils.h b/paddle/math/tests/TestUtils.h index 2edb07de01..5f9fab7245 100644 --- a/paddle/math/tests/TestUtils.h +++ b/paddle/math/tests/TestUtils.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_Allocator.cpp b/paddle/math/tests/test_Allocator.cpp index 084322a1ca..440fcda0fe 100644 --- a/paddle/math/tests/test_Allocator.cpp +++ b/paddle/math/tests/test_Allocator.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_BaseMatrix.cpp b/paddle/math/tests/test_BaseMatrix.cpp index f8c795a639..a4683918ca 100644 --- a/paddle/math/tests/test_BaseMatrix.cpp +++ b/paddle/math/tests/test_BaseMatrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_CpuGpuVector.cpp b/paddle/math/tests/test_CpuGpuVector.cpp index 7b50b020cd..c671735875 100644 --- a/paddle/math/tests/test_CpuGpuVector.cpp +++ b/paddle/math/tests/test_CpuGpuVector.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_ExecViaCpu.cpp b/paddle/math/tests/test_ExecViaCpu.cpp index b3eca19a72..b328ebf554 100644 --- a/paddle/math/tests/test_ExecViaCpu.cpp +++ b/paddle/math/tests/test_ExecViaCpu.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_FPException.cpp b/paddle/math/tests/test_FPException.cpp index f996e0dadd..6aa5891bce 100644 --- a/paddle/math/tests/test_FPException.cpp +++ b/paddle/math/tests/test_FPException.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_GpuProfiler.cpp b/paddle/math/tests/test_GpuProfiler.cpp index c3542b7834..e5fd6f4523 100644 --- a/paddle/math/tests/test_GpuProfiler.cpp +++ b/paddle/math/tests/test_GpuProfiler.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_Matrix.cpp b/paddle/math/tests/test_Matrix.cpp index edc9d74103..adb5fbd9fa 100644 --- a/paddle/math/tests/test_Matrix.cpp +++ b/paddle/math/tests/test_Matrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_SIMDFunctions.cpp b/paddle/math/tests/test_SIMDFunctions.cpp index 8405b96fc2..2c54121d99 100644 --- a/paddle/math/tests/test_SIMDFunctions.cpp +++ b/paddle/math/tests/test_SIMDFunctions.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_SparseMatrix.cpp b/paddle/math/tests/test_SparseMatrix.cpp index 3788218aab..88b75b6d83 100644 --- a/paddle/math/tests/test_SparseMatrix.cpp +++ b/paddle/math/tests/test_SparseMatrix.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_Tensor.cu b/paddle/math/tests/test_Tensor.cu index 8fa402055a..1859b9fc13 100644 --- a/paddle/math/tests/test_Tensor.cu +++ b/paddle/math/tests/test_Tensor.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_TrainingAlgorithm.cpp b/paddle/math/tests/test_TrainingAlgorithm.cpp index b40c8d9dae..93a930cc2f 100644 --- a/paddle/math/tests/test_TrainingAlgorithm.cpp +++ b/paddle/math/tests/test_TrainingAlgorithm.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_batchTranspose.cpp b/paddle/math/tests/test_batchTranspose.cpp index a9596992b2..88631c62b8 100644 --- a/paddle/math/tests/test_batchTranspose.cpp +++ b/paddle/math/tests/test_batchTranspose.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_lazyAssign.cu b/paddle/math/tests/test_lazyAssign.cu index 52dfdacffe..16541edb54 100644 --- a/paddle/math/tests/test_lazyAssign.cu +++ b/paddle/math/tests/test_lazyAssign.cu @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_matrixCompare.cpp b/paddle/math/tests/test_matrixCompare.cpp index 0883066947..713792d82b 100644 --- a/paddle/math/tests/test_matrixCompare.cpp +++ b/paddle/math/tests/test_matrixCompare.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_matrixUtil.h b/paddle/math/tests/test_matrixUtil.h index 5300e7168b..9aa74b1519 100644 --- a/paddle/math/tests/test_matrixUtil.h +++ b/paddle/math/tests/test_matrixUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_perturbation.cpp b/paddle/math/tests/test_perturbation.cpp index 837c2f47ba..eaf4dfea66 100644 --- a/paddle/math/tests/test_perturbation.cpp +++ b/paddle/math/tests/test_perturbation.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/math/tests/test_sparseMatrixCompare.cpp b/paddle/math/tests/test_sparseMatrixCompare.cpp index d7aa20eb98..eff2c502bb 100644 --- a/paddle/math/tests/test_sparseMatrixCompare.cpp +++ b/paddle/math/tests/test_sparseMatrixCompare.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Argument.cpp b/paddle/parameter/Argument.cpp index 0f414b4463..b632a11bbd 100644 --- a/paddle/parameter/Argument.cpp +++ b/paddle/parameter/Argument.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Argument.h b/paddle/parameter/Argument.h index 2b20122deb..69d57a28c0 100644 --- a/paddle/parameter/Argument.h +++ b/paddle/parameter/Argument.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/AverageOptimizer.cpp b/paddle/parameter/AverageOptimizer.cpp index 593594761e..e51ca56520 100644 --- a/paddle/parameter/AverageOptimizer.cpp +++ b/paddle/parameter/AverageOptimizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/AverageOptimizer.h b/paddle/parameter/AverageOptimizer.h index ccc2612608..9fd3f75baa 100644 --- a/paddle/parameter/AverageOptimizer.h +++ b/paddle/parameter/AverageOptimizer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/FirstOrderOptimizer.cpp b/paddle/parameter/FirstOrderOptimizer.cpp index 9e363fb20d..17268d3715 100644 --- a/paddle/parameter/FirstOrderOptimizer.cpp +++ b/paddle/parameter/FirstOrderOptimizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/FirstOrderOptimizer.h b/paddle/parameter/FirstOrderOptimizer.h index a9a2ffdd41..095019b74f 100644 --- a/paddle/parameter/FirstOrderOptimizer.h +++ b/paddle/parameter/FirstOrderOptimizer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/LearningRateScheduler.cpp b/paddle/parameter/LearningRateScheduler.cpp index a7412500cc..66448b2c5f 100644 --- a/paddle/parameter/LearningRateScheduler.cpp +++ b/paddle/parameter/LearningRateScheduler.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/LearningRateScheduler.h b/paddle/parameter/LearningRateScheduler.h index e987c3dcde..53b9dba446 100644 --- a/paddle/parameter/LearningRateScheduler.h +++ b/paddle/parameter/LearningRateScheduler.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/OptimizerFunctions.cpp b/paddle/parameter/OptimizerFunctions.cpp index 6fd7964347..a4af1b4705 100644 --- a/paddle/parameter/OptimizerFunctions.cpp +++ b/paddle/parameter/OptimizerFunctions.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/OptimizerFunctions.h b/paddle/parameter/OptimizerFunctions.h index a5f8b2c569..4f7370b6ba 100644 --- a/paddle/parameter/OptimizerFunctions.h +++ b/paddle/parameter/OptimizerFunctions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/OptimizerWithRegularizer.cpp b/paddle/parameter/OptimizerWithRegularizer.cpp index 5381e7bef3..85f13c8bc0 100644 --- a/paddle/parameter/OptimizerWithRegularizer.cpp +++ b/paddle/parameter/OptimizerWithRegularizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/OptimizerWithRegularizer.h b/paddle/parameter/OptimizerWithRegularizer.h index ebe23c7397..0e1c444d28 100644 --- a/paddle/parameter/OptimizerWithRegularizer.h +++ b/paddle/parameter/OptimizerWithRegularizer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParallelParameter.cpp b/paddle/parameter/ParallelParameter.cpp index 99b20a59ca..b3182306a4 100644 --- a/paddle/parameter/ParallelParameter.cpp +++ b/paddle/parameter/ParallelParameter.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParallelParameter.h b/paddle/parameter/ParallelParameter.h index 2b65321fe2..b0fe82d3c4 100644 --- a/paddle/parameter/ParallelParameter.h +++ b/paddle/parameter/ParallelParameter.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Parameter.cpp b/paddle/parameter/Parameter.cpp index 7e37bf225b..3b06650e0c 100644 --- a/paddle/parameter/Parameter.cpp +++ b/paddle/parameter/Parameter.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Parameter.h b/paddle/parameter/Parameter.h index 1c159d669a..6b0600517a 100644 --- a/paddle/parameter/Parameter.h +++ b/paddle/parameter/Parameter.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterOptimizer.cpp b/paddle/parameter/ParameterOptimizer.cpp index 2a71d6aee4..7c8c6978e2 100644 --- a/paddle/parameter/ParameterOptimizer.cpp +++ b/paddle/parameter/ParameterOptimizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterOptimizer.h b/paddle/parameter/ParameterOptimizer.h index 21a148333c..2bdc793d60 100644 --- a/paddle/parameter/ParameterOptimizer.h +++ b/paddle/parameter/ParameterOptimizer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdateFunctions.cpp b/paddle/parameter/ParameterUpdateFunctions.cpp index 510ec5bf48..c8af7105c7 100644 --- a/paddle/parameter/ParameterUpdateFunctions.cpp +++ b/paddle/parameter/ParameterUpdateFunctions.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdateFunctions.h b/paddle/parameter/ParameterUpdateFunctions.h index 2d98030bd2..7374843d80 100644 --- a/paddle/parameter/ParameterUpdateFunctions.h +++ b/paddle/parameter/ParameterUpdateFunctions.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdaterBase.cpp b/paddle/parameter/ParameterUpdaterBase.cpp index e706742053..b938270ce1 100644 --- a/paddle/parameter/ParameterUpdaterBase.cpp +++ b/paddle/parameter/ParameterUpdaterBase.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdaterBase.h b/paddle/parameter/ParameterUpdaterBase.h index ffd2980261..5401046f67 100644 --- a/paddle/parameter/ParameterUpdaterBase.h +++ b/paddle/parameter/ParameterUpdaterBase.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdaterHook.cpp b/paddle/parameter/ParameterUpdaterHook.cpp index 7d85a32c0c..466560c437 100644 --- a/paddle/parameter/ParameterUpdaterHook.cpp +++ b/paddle/parameter/ParameterUpdaterHook.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/ParameterUpdaterHook.h b/paddle/parameter/ParameterUpdaterHook.h index 553282bcaa..1f4506441d 100644 --- a/paddle/parameter/ParameterUpdaterHook.h +++ b/paddle/parameter/ParameterUpdaterHook.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Regularizer.cpp b/paddle/parameter/Regularizer.cpp index a9bddc1596..4420ee0031 100644 --- a/paddle/parameter/Regularizer.cpp +++ b/paddle/parameter/Regularizer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Regularizer.h b/paddle/parameter/Regularizer.h index 5baaccc00d..6d54773098 100644 --- a/paddle/parameter/Regularizer.h +++ b/paddle/parameter/Regularizer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Weight.cpp b/paddle/parameter/Weight.cpp index c138010607..f366a2b53f 100644 --- a/paddle/parameter/Weight.cpp +++ b/paddle/parameter/Weight.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/Weight.h b/paddle/parameter/Weight.h index 531b571cbc..6e7a49154e 100644 --- a/paddle/parameter/Weight.h +++ b/paddle/parameter/Weight.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/parameter/tests/test_common.cpp b/paddle/parameter/tests/test_common.cpp index 1a64fe3352..4e4d0ccfa2 100644 --- a/paddle/parameter/tests/test_common.cpp +++ b/paddle/parameter/tests/test_common.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/BaseClient.cpp b/paddle/pserver/BaseClient.cpp index ff83970ab1..62fafc1891 100644 --- a/paddle/pserver/BaseClient.cpp +++ b/paddle/pserver/BaseClient.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/BaseClient.h b/paddle/pserver/BaseClient.h index 3a501172b7..5924f80684 100644 --- a/paddle/pserver/BaseClient.h +++ b/paddle/pserver/BaseClient.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/LightNetwork.cpp b/paddle/pserver/LightNetwork.cpp index 1830170a16..9a398d4f45 100644 --- a/paddle/pserver/LightNetwork.cpp +++ b/paddle/pserver/LightNetwork.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/LightNetwork.h b/paddle/pserver/LightNetwork.h index b7d7bc7902..7aff007a27 100644 --- a/paddle/pserver/LightNetwork.h +++ b/paddle/pserver/LightNetwork.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ParameterClient2.cpp b/paddle/pserver/ParameterClient2.cpp index 28cc0ae2dd..31418822b3 100644 --- a/paddle/pserver/ParameterClient2.cpp +++ b/paddle/pserver/ParameterClient2.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ParameterClient2.h b/paddle/pserver/ParameterClient2.h index af8dd41ec4..0f180722e3 100644 --- a/paddle/pserver/ParameterClient2.h +++ b/paddle/pserver/ParameterClient2.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ParameterServer2.cpp b/paddle/pserver/ParameterServer2.cpp index b7f999f8b1..ac70efc64f 100644 --- a/paddle/pserver/ParameterServer2.cpp +++ b/paddle/pserver/ParameterServer2.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ParameterServer2.h b/paddle/pserver/ParameterServer2.h index ccaea42e7d..47122f3632 100644 --- a/paddle/pserver/ParameterServer2.h +++ b/paddle/pserver/ParameterServer2.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ParameterServer2Main.cpp b/paddle/pserver/ParameterServer2Main.cpp index b15ef8c3cc..1ba9b48c23 100644 --- a/paddle/pserver/ParameterServer2Main.cpp +++ b/paddle/pserver/ParameterServer2Main.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ProtoServer.cpp b/paddle/pserver/ProtoServer.cpp index 2f6d911a01..410317ece2 100644 --- a/paddle/pserver/ProtoServer.cpp +++ b/paddle/pserver/ProtoServer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/ProtoServer.h b/paddle/pserver/ProtoServer.h index cf08e24ff3..97b7bf167d 100644 --- a/paddle/pserver/ProtoServer.h +++ b/paddle/pserver/ProtoServer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/RDMANetwork.h b/paddle/pserver/RDMANetwork.h index 4e492a3afd..caef65134b 100644 --- a/paddle/pserver/RDMANetwork.h +++ b/paddle/pserver/RDMANetwork.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/SocketChannel.cpp b/paddle/pserver/SocketChannel.cpp index 4ebc47d326..f3e74257f6 100644 --- a/paddle/pserver/SocketChannel.cpp +++ b/paddle/pserver/SocketChannel.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/SocketChannel.h b/paddle/pserver/SocketChannel.h index 472b37a122..6c3dd20d7b 100644 --- a/paddle/pserver/SocketChannel.h +++ b/paddle/pserver/SocketChannel.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/SparseParameterDistribution.cpp b/paddle/pserver/SparseParameterDistribution.cpp index 2085b22a95..0068f85b52 100644 --- a/paddle/pserver/SparseParameterDistribution.cpp +++ b/paddle/pserver/SparseParameterDistribution.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/SparseParameterDistribution.h b/paddle/pserver/SparseParameterDistribution.h index af2b43af0f..dc63b065a7 100644 --- a/paddle/pserver/SparseParameterDistribution.h +++ b/paddle/pserver/SparseParameterDistribution.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/test/SocketTest.cpp b/paddle/pserver/test/SocketTest.cpp index 24c90f1078..528f5e381e 100644 --- a/paddle/pserver/test/SocketTest.cpp +++ b/paddle/pserver/test/SocketTest.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/test/test_ParameterServer2.cpp b/paddle/pserver/test/test_ParameterServer2.cpp index eb813e92d6..493b6d060c 100644 --- a/paddle/pserver/test/test_ParameterServer2.cpp +++ b/paddle/pserver/test/test_ParameterServer2.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/test/test_ProtoServer.cpp b/paddle/pserver/test/test_ProtoServer.cpp index 79d1f2743a..cfed0d30d3 100644 --- a/paddle/pserver/test/test_ProtoServer.cpp +++ b/paddle/pserver/test/test_ProtoServer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/pserver/test/test_ProtoServer.sh b/paddle/pserver/test/test_ProtoServer.sh index a87b1b1ddc..970c90b494 100755 --- a/paddle/pserver/test/test_ProtoServer.sh +++ b/paddle/pserver/test/test_ProtoServer.sh @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/py_paddle/__init__.py b/paddle/py_paddle/__init__.py index f8399f9c63..5504d1d50c 100644 --- a/paddle/py_paddle/__init__.py +++ b/paddle/py_paddle/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/py_paddle/dataprovider_converter.py b/paddle/py_paddle/dataprovider_converter.py index d64c7b20cb..edcefba6a8 100644 --- a/paddle/py_paddle/dataprovider_converter.py +++ b/paddle/py_paddle/dataprovider_converter.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/py_paddle/util.py b/paddle/py_paddle/util.py index 35a355ef29..d6bbf9a5a9 100644 --- a/paddle/py_paddle/util.py +++ b/paddle/py_paddle/util.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/scripts/cluster_train/conf.py b/paddle/scripts/cluster_train/conf.py index f1114a5920..c77d7584d3 100644 --- a/paddle/scripts/cluster_train/conf.py +++ b/paddle/scripts/cluster_train/conf.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/scripts/cluster_train/paddle.py b/paddle/scripts/cluster_train/paddle.py index 7343a600c1..9b03ed1d8f 100644 --- a/paddle/scripts/cluster_train/paddle.py +++ b/paddle/scripts/cluster_train/paddle.py @@ -1,5 +1,5 @@ #!/usr/bin/python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/setup.py.in b/paddle/setup.py.in index 1a15eafd55..b4c38a41b8 100644 --- a/paddle/setup.py.in +++ b/paddle/setup.py.in @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/MergeModel.cpp b/paddle/trainer/MergeModel.cpp index 1d15c66d4d..8cb2873feb 100644 --- a/paddle/trainer/MergeModel.cpp +++ b/paddle/trainer/MergeModel.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ParamUtil.cpp b/paddle/trainer/ParamUtil.cpp index 2be9cd6223..200417ebfc 100644 --- a/paddle/trainer/ParamUtil.cpp +++ b/paddle/trainer/ParamUtil.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ParamUtil.h b/paddle/trainer/ParamUtil.h index 3923941c3d..8fa6fda75c 100644 --- a/paddle/trainer/ParamUtil.h +++ b/paddle/trainer/ParamUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ParameterUpdater.cpp b/paddle/trainer/ParameterUpdater.cpp index 6001a0b391..8b5b95da5b 100644 --- a/paddle/trainer/ParameterUpdater.cpp +++ b/paddle/trainer/ParameterUpdater.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ParameterUpdater.h b/paddle/trainer/ParameterUpdater.h index b83b4cf55e..81ac374425 100644 --- a/paddle/trainer/ParameterUpdater.h +++ b/paddle/trainer/ParameterUpdater.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/RemoteParameterUpdater.cpp b/paddle/trainer/RemoteParameterUpdater.cpp index d83bb5b10a..702ea07f8a 100644 --- a/paddle/trainer/RemoteParameterUpdater.cpp +++ b/paddle/trainer/RemoteParameterUpdater.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/RemoteParameterUpdater.h b/paddle/trainer/RemoteParameterUpdater.h index a40884724c..46ce4be146 100644 --- a/paddle/trainer/RemoteParameterUpdater.h +++ b/paddle/trainer/RemoteParameterUpdater.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/Tester.cpp b/paddle/trainer/Tester.cpp index 6a5b7241a0..97d1b53934 100644 --- a/paddle/trainer/Tester.cpp +++ b/paddle/trainer/Tester.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/Tester.h b/paddle/trainer/Tester.h index a9de9fe208..ae7e0e93bf 100644 --- a/paddle/trainer/Tester.h +++ b/paddle/trainer/Tester.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TesterConfig.h b/paddle/trainer/TesterConfig.h index f490e57344..9ff145a8a1 100644 --- a/paddle/trainer/TesterConfig.h +++ b/paddle/trainer/TesterConfig.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ThreadParameterUpdater.cpp b/paddle/trainer/ThreadParameterUpdater.cpp index cc22851d8e..bee7f061fe 100644 --- a/paddle/trainer/ThreadParameterUpdater.cpp +++ b/paddle/trainer/ThreadParameterUpdater.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/ThreadParameterUpdater.h b/paddle/trainer/ThreadParameterUpdater.h index 5a5e3f1d4b..492692dbe5 100644 --- a/paddle/trainer/ThreadParameterUpdater.h +++ b/paddle/trainer/ThreadParameterUpdater.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/Trainer.cpp b/paddle/trainer/Trainer.cpp index 9c83c207ed..85610ec04e 100644 --- a/paddle/trainer/Trainer.cpp +++ b/paddle/trainer/Trainer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/Trainer.h b/paddle/trainer/Trainer.h index 899607c7c0..f50b56143d 100644 --- a/paddle/trainer/Trainer.h +++ b/paddle/trainer/Trainer.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerBenchmark.cpp b/paddle/trainer/TrainerBenchmark.cpp index 54862e95b4..5c3177c808 100644 --- a/paddle/trainer/TrainerBenchmark.cpp +++ b/paddle/trainer/TrainerBenchmark.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerConfigHelper.cpp b/paddle/trainer/TrainerConfigHelper.cpp index ee5b1e0a9c..2017a08d20 100644 --- a/paddle/trainer/TrainerConfigHelper.cpp +++ b/paddle/trainer/TrainerConfigHelper.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerConfigHelper.h b/paddle/trainer/TrainerConfigHelper.h index d206849641..2c5c492ce8 100644 --- a/paddle/trainer/TrainerConfigHelper.h +++ b/paddle/trainer/TrainerConfigHelper.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerInternal.cpp b/paddle/trainer/TrainerInternal.cpp index b1c3bf26d2..1b49d4aa28 100644 --- a/paddle/trainer/TrainerInternal.cpp +++ b/paddle/trainer/TrainerInternal.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerInternal.h b/paddle/trainer/TrainerInternal.h index 962d53a30e..b67711a721 100644 --- a/paddle/trainer/TrainerInternal.h +++ b/paddle/trainer/TrainerInternal.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerInternalConfig.cpp b/paddle/trainer/TrainerInternalConfig.cpp index 0dc74cb3b3..a017cdec9d 100644 --- a/paddle/trainer/TrainerInternalConfig.cpp +++ b/paddle/trainer/TrainerInternalConfig.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerInternalConfig.h b/paddle/trainer/TrainerInternalConfig.h index b7bfd29abd..fd6fdf45e6 100644 --- a/paddle/trainer/TrainerInternalConfig.h +++ b/paddle/trainer/TrainerInternalConfig.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/TrainerMain.cpp b/paddle/trainer/TrainerMain.cpp index e23e745d99..7a18f9836c 100644 --- a/paddle/trainer/TrainerMain.cpp +++ b/paddle/trainer/TrainerMain.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/__init__.py b/paddle/trainer/tests/__init__.py index c90af2ee00..f662d68263 100644 --- a/paddle/trainer/tests/__init__.py +++ b/paddle/trainer/tests/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/chunking.conf b/paddle/trainer/tests/chunking.conf index 01c15fab5f..d88df919df 100644 --- a/paddle/trainer/tests/chunking.conf +++ b/paddle/trainer/tests/chunking.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/config_parser_test.py b/paddle/trainer/tests/config_parser_test.py index c5ec315d6b..db66ebb5b7 100644 --- a/paddle/trainer/tests/config_parser_test.py +++ b/paddle/trainer/tests/config_parser_test.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/gen_proto_data.py b/paddle/trainer/tests/gen_proto_data.py index a3dbc10c88..8cc6d44673 100644 --- a/paddle/trainer/tests/gen_proto_data.py +++ b/paddle/trainer/tests/gen_proto_data.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config.conf b/paddle/trainer/tests/sample_trainer_config.conf index 15901065b2..2697832840 100644 --- a/paddle/trainer/tests/sample_trainer_config.conf +++ b/paddle/trainer/tests/sample_trainer_config.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_hsigmoid.conf b/paddle/trainer/tests/sample_trainer_config_hsigmoid.conf index 174cb5e25f..e4abe31d48 100644 --- a/paddle/trainer/tests/sample_trainer_config_hsigmoid.conf +++ b/paddle/trainer/tests/sample_trainer_config_hsigmoid.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_opt_a.conf b/paddle/trainer/tests/sample_trainer_config_opt_a.conf index f5b1988dda..b1744db8d6 100644 --- a/paddle/trainer/tests/sample_trainer_config_opt_a.conf +++ b/paddle/trainer/tests/sample_trainer_config_opt_a.conf @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_opt_b.conf b/paddle/trainer/tests/sample_trainer_config_opt_b.conf index f5b1988dda..b1744db8d6 100644 --- a/paddle/trainer/tests/sample_trainer_config_opt_b.conf +++ b/paddle/trainer/tests/sample_trainer_config_opt_b.conf @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_parallel.conf b/paddle/trainer/tests/sample_trainer_config_parallel.conf index e35a1f26da..e2b8b3ecda 100644 --- a/paddle/trainer/tests/sample_trainer_config_parallel.conf +++ b/paddle/trainer/tests/sample_trainer_config_parallel.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_qb_rnn.conf b/paddle/trainer/tests/sample_trainer_config_qb_rnn.conf index d254cc5700..d19222360c 100644 --- a/paddle/trainer/tests/sample_trainer_config_qb_rnn.conf +++ b/paddle/trainer/tests/sample_trainer_config_qb_rnn.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_config_rnn.conf b/paddle/trainer/tests/sample_trainer_config_rnn.conf index cbb6663029..b720d4d5a6 100644 --- a/paddle/trainer/tests/sample_trainer_config_rnn.conf +++ b/paddle/trainer/tests/sample_trainer_config_rnn.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_nest_rnn_gen.conf b/paddle/trainer/tests/sample_trainer_nest_rnn_gen.conf index 613fd325e1..d669fbc40c 100644 --- a/paddle/trainer/tests/sample_trainer_nest_rnn_gen.conf +++ b/paddle/trainer/tests/sample_trainer_nest_rnn_gen.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/sample_trainer_rnn_gen.conf b/paddle/trainer/tests/sample_trainer_rnn_gen.conf index ec1c12cc89..2b337282f6 100644 --- a/paddle/trainer/tests/sample_trainer_rnn_gen.conf +++ b/paddle/trainer/tests/sample_trainer_rnn_gen.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/testPyDataWrapper.py b/paddle/trainer/tests/testPyDataWrapper.py index 4607bec24e..2c29a27433 100644 --- a/paddle/trainer/tests/testPyDataWrapper.py +++ b/paddle/trainer/tests/testPyDataWrapper.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_Compare.cpp b/paddle/trainer/tests/test_Compare.cpp index 03312f9e47..07a47b2990 100644 --- a/paddle/trainer/tests/test_Compare.cpp +++ b/paddle/trainer/tests/test_Compare.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_CompareSparse.cpp b/paddle/trainer/tests/test_CompareSparse.cpp index a7c6862ce3..3fea3a3c24 100644 --- a/paddle/trainer/tests/test_CompareSparse.cpp +++ b/paddle/trainer/tests/test_CompareSparse.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_CompareTwoNets.cpp b/paddle/trainer/tests/test_CompareTwoNets.cpp index 81320da6ac..7e5449dcba 100644 --- a/paddle/trainer/tests/test_CompareTwoNets.cpp +++ b/paddle/trainer/tests/test_CompareTwoNets.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_CompareTwoOpts.cpp b/paddle/trainer/tests/test_CompareTwoOpts.cpp index a52f2fa7e7..4d051b537c 100644 --- a/paddle/trainer/tests/test_CompareTwoOpts.cpp +++ b/paddle/trainer/tests/test_CompareTwoOpts.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_Prediction.cpp b/paddle/trainer/tests/test_Prediction.cpp index 6db33439b3..322121a579 100644 --- a/paddle/trainer/tests/test_Prediction.cpp +++ b/paddle/trainer/tests/test_Prediction.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_PyDataProviderWrapper.cpp b/paddle/trainer/tests/test_PyDataProviderWrapper.cpp index e53291386c..5c5c6d5346 100644 --- a/paddle/trainer/tests/test_PyDataProviderWrapper.cpp +++ b/paddle/trainer/tests/test_PyDataProviderWrapper.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_Trainer.cpp b/paddle/trainer/tests/test_Trainer.cpp index 900c05af85..0fede59f8d 100644 --- a/paddle/trainer/tests/test_Trainer.cpp +++ b/paddle/trainer/tests/test_Trainer.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_TrainerOnePass.cpp b/paddle/trainer/tests/test_TrainerOnePass.cpp index da2954d166..1d9dce1b0e 100644 --- a/paddle/trainer/tests/test_TrainerOnePass.cpp +++ b/paddle/trainer/tests/test_TrainerOnePass.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_config.conf b/paddle/trainer/tests/test_config.conf index 2a4548896f..d1bb9b877f 100644 --- a/paddle/trainer/tests/test_config.conf +++ b/paddle/trainer/tests/test_config.conf @@ -1,5 +1,5 @@ #edit-mode: -*- python -*- -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/paddle/trainer/tests/test_recurrent_machine_generation.cpp b/paddle/trainer/tests/test_recurrent_machine_generation.cpp index 49e8a97ad0..b52acc2ca7 100644 --- a/paddle/trainer/tests/test_recurrent_machine_generation.cpp +++ b/paddle/trainer/tests/test_recurrent_machine_generation.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/BarrierStat.cpp b/paddle/utils/BarrierStat.cpp index 82c5b84e59..5040deefd0 100644 --- a/paddle/utils/BarrierStat.cpp +++ b/paddle/utils/BarrierStat.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/BarrierStat.h b/paddle/utils/BarrierStat.h index 661340ad27..3c5c0885d6 100644 --- a/paddle/utils/BarrierStat.h +++ b/paddle/utils/BarrierStat.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/ClassRegistrar.h b/paddle/utils/ClassRegistrar.h index ee58ccb2ad..1ac27bafab 100644 --- a/paddle/utils/ClassRegistrar.h +++ b/paddle/utils/ClassRegistrar.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/CommandLineParser.cpp b/paddle/utils/CommandLineParser.cpp index 307e304bb0..14f83241c5 100644 --- a/paddle/utils/CommandLineParser.cpp +++ b/paddle/utils/CommandLineParser.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/CommandLineParser.h b/paddle/utils/CommandLineParser.h index c46567913e..3d25bc3b0b 100644 --- a/paddle/utils/CommandLineParser.h +++ b/paddle/utils/CommandLineParser.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/CompilerMacros.h b/paddle/utils/CompilerMacros.h index 4236d750c4..e50093f7fc 100644 --- a/paddle/utils/CompilerMacros.h +++ b/paddle/utils/CompilerMacros.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/CustomStackTrace.cpp b/paddle/utils/CustomStackTrace.cpp index 8740fe662e..730788cb98 100644 --- a/paddle/utils/CustomStackTrace.cpp +++ b/paddle/utils/CustomStackTrace.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/CustomStackTrace.h b/paddle/utils/CustomStackTrace.h index 878e14eb5f..5686f3c84c 100644 --- a/paddle/utils/CustomStackTrace.h +++ b/paddle/utils/CustomStackTrace.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/DisableCopy.h b/paddle/utils/DisableCopy.h index e991c07cdf..41de98bbde 100644 --- a/paddle/utils/DisableCopy.h +++ b/paddle/utils/DisableCopy.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Excepts.cpp b/paddle/utils/Excepts.cpp index b2fad3ac9d..4ddce35ed3 100644 --- a/paddle/utils/Excepts.cpp +++ b/paddle/utils/Excepts.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Excepts.h b/paddle/utils/Excepts.h index a84a2d33a6..dc3369b7e8 100644 --- a/paddle/utils/Excepts.h +++ b/paddle/utils/Excepts.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Flags.cpp b/paddle/utils/Flags.cpp index 6fae24e1b5..1c9e602f45 100644 --- a/paddle/utils/Flags.cpp +++ b/paddle/utils/Flags.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Flags.h b/paddle/utils/Flags.h index dda60c3f96..922533d63e 100644 --- a/paddle/utils/Flags.h +++ b/paddle/utils/Flags.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/GlobalConstants.cpp b/paddle/utils/GlobalConstants.cpp index d769cd1ee7..9e8dade0b2 100644 --- a/paddle/utils/GlobalConstants.cpp +++ b/paddle/utils/GlobalConstants.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/GlobalConstants.h b/paddle/utils/GlobalConstants.h index 4c74c17a50..707346f2c7 100644 --- a/paddle/utils/GlobalConstants.h +++ b/paddle/utils/GlobalConstants.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Locks.h b/paddle/utils/Locks.h index 5990e16570..0f922f3548 100644 --- a/paddle/utils/Locks.h +++ b/paddle/utils/Locks.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Logging.cpp b/paddle/utils/Logging.cpp index 14303bd4c7..3c31633e58 100644 --- a/paddle/utils/Logging.cpp +++ b/paddle/utils/Logging.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Logging.h b/paddle/utils/Logging.h index e9029b421f..c91ca9fecc 100644 --- a/paddle/utils/Logging.h +++ b/paddle/utils/Logging.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/PythonUtil.cpp b/paddle/utils/PythonUtil.cpp index 7f17a82522..a9c6a20997 100644 --- a/paddle/utils/PythonUtil.cpp +++ b/paddle/utils/PythonUtil.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/PythonUtil.h b/paddle/utils/PythonUtil.h index 65677d9010..2cbc2fdd37 100644 --- a/paddle/utils/PythonUtil.h +++ b/paddle/utils/PythonUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Queue.h b/paddle/utils/Queue.h index 58d17e86c4..37748345a4 100644 --- a/paddle/utils/Queue.h +++ b/paddle/utils/Queue.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Stat.cpp b/paddle/utils/Stat.cpp index ab140c3350..01ea535cfd 100644 --- a/paddle/utils/Stat.cpp +++ b/paddle/utils/Stat.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Stat.h b/paddle/utils/Stat.h index 1ef688ea8d..9be79e8859 100644 --- a/paddle/utils/Stat.h +++ b/paddle/utils/Stat.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/StringUtil.cpp b/paddle/utils/StringUtil.cpp index b416cda4af..0c98e6db34 100644 --- a/paddle/utils/StringUtil.cpp +++ b/paddle/utils/StringUtil.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/StringUtil.h b/paddle/utils/StringUtil.h index 8b44dad192..8a63ca23b4 100644 --- a/paddle/utils/StringUtil.h +++ b/paddle/utils/StringUtil.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Thread.h b/paddle/utils/Thread.h index ade0ee496f..435dff2f66 100644 --- a/paddle/utils/Thread.h +++ b/paddle/utils/Thread.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/ThreadLocal.cpp b/paddle/utils/ThreadLocal.cpp index 49d4b15265..c9b32784d9 100644 --- a/paddle/utils/ThreadLocal.cpp +++ b/paddle/utils/ThreadLocal.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/ThreadLocal.h b/paddle/utils/ThreadLocal.h index 06c8b392af..b6e31bd05b 100644 --- a/paddle/utils/ThreadLocal.h +++ b/paddle/utils/ThreadLocal.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/TypeDefs.h b/paddle/utils/TypeDefs.h index e8be779bea..c50a05e82d 100644 --- a/paddle/utils/TypeDefs.h +++ b/paddle/utils/TypeDefs.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Util.cpp b/paddle/utils/Util.cpp index bc727cfa74..f48726bff0 100644 --- a/paddle/utils/Util.cpp +++ b/paddle/utils/Util.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Util.h b/paddle/utils/Util.h index ed38f8fa60..ff67439da6 100644 --- a/paddle/utils/Util.h +++ b/paddle/utils/Util.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Version.cpp b/paddle/utils/Version.cpp index e706983918..88fb333f79 100644 --- a/paddle/utils/Version.cpp +++ b/paddle/utils/Version.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/Version.h b/paddle/utils/Version.h index e6c799644e..ac04963c2c 100644 --- a/paddle/utils/Version.h +++ b/paddle/utils/Version.h @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/arch/linux/Locks.cpp b/paddle/utils/arch/linux/Locks.cpp index 93016daeae..2a6f96e04d 100644 --- a/paddle/utils/arch/linux/Locks.cpp +++ b/paddle/utils/arch/linux/Locks.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/arch/osx/Locks.cpp b/paddle/utils/arch/osx/Locks.cpp index ae563a6afd..8590226431 100644 --- a/paddle/utils/arch/osx/Locks.cpp +++ b/paddle/utils/arch/osx/Locks.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_CommandLineParser.cpp b/paddle/utils/tests/test_CommandLineParser.cpp index 5ecfb2b4f5..9a1d2391a8 100644 --- a/paddle/utils/tests/test_CommandLineParser.cpp +++ b/paddle/utils/tests/test_CommandLineParser.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_CustomStackTrace.cpp b/paddle/utils/tests/test_CustomStackTrace.cpp index 3bfb381ed9..512330b49e 100644 --- a/paddle/utils/tests/test_CustomStackTrace.cpp +++ b/paddle/utils/tests/test_CustomStackTrace.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_CustomStackTracePrint.cpp b/paddle/utils/tests/test_CustomStackTracePrint.cpp index d39a190961..60ba210b70 100644 --- a/paddle/utils/tests/test_CustomStackTracePrint.cpp +++ b/paddle/utils/tests/test_CustomStackTracePrint.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_Logging.cpp b/paddle/utils/tests/test_Logging.cpp index 9f477fab14..667864aa75 100644 --- a/paddle/utils/tests/test_Logging.cpp +++ b/paddle/utils/tests/test_Logging.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_SpinLock.cpp b/paddle/utils/tests/test_SpinLock.cpp index 77d281962c..9c7ad05b0b 100644 --- a/paddle/utils/tests/test_SpinLock.cpp +++ b/paddle/utils/tests/test_SpinLock.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_StringUtils.cpp b/paddle/utils/tests/test_StringUtils.cpp index 2c699b791f..fdc914d1bc 100644 --- a/paddle/utils/tests/test_StringUtils.cpp +++ b/paddle/utils/tests/test_StringUtils.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_Thread.cpp b/paddle/utils/tests/test_Thread.cpp index 154db5d9c6..b069be1d7a 100644 --- a/paddle/utils/tests/test_Thread.cpp +++ b/paddle/utils/tests/test_Thread.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/paddle/utils/tests/test_ThreadBarrier.cpp b/paddle/utils/tests/test_ThreadBarrier.cpp index 20b9babd94..997a393683 100644 --- a/paddle/utils/tests/test_ThreadBarrier.cpp +++ b/paddle/utils/tests/test_ThreadBarrier.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/DataConfig.proto.m4 b/proto/DataConfig.proto.m4 index 01d451ff7d..1f8e3f4f3e 100644 --- a/proto/DataConfig.proto.m4 +++ b/proto/DataConfig.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/DataFormat.proto.m4 b/proto/DataFormat.proto.m4 index 8a4a0be1b3..54e9fd008e 100644 --- a/proto/DataFormat.proto.m4 +++ b/proto/DataFormat.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/ModelConfig.proto.m4 b/proto/ModelConfig.proto.m4 index 4e8ed36f4e..ccad69a3c2 100644 --- a/proto/ModelConfig.proto.m4 +++ b/proto/ModelConfig.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/ParameterConfig.proto.m4 b/proto/ParameterConfig.proto.m4 index 26e7c3ef77..b5c0fea6c3 100644 --- a/proto/ParameterConfig.proto.m4 +++ b/proto/ParameterConfig.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/ParameterService.proto.m4 b/proto/ParameterService.proto.m4 index 0b3f14a2ee..25b0991583 100644 --- a/proto/ParameterService.proto.m4 +++ b/proto/ParameterService.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/proto/TrainerConfig.proto.m4 b/proto/TrainerConfig.proto.m4 index 965c9cd393..4684203b03 100644 --- a/proto/TrainerConfig.proto.m4 +++ b/proto/TrainerConfig.proto.m4 @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/python/paddle/__init__.py b/python/paddle/__init__.py index c90af2ee00..f662d68263 100644 --- a/python/paddle/__init__.py +++ b/python/paddle/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/proto/__init__.py b/python/paddle/proto/__init__.py index cd6a59ecbb..07406a841e 100644 --- a/python/paddle/proto/__init__.py +++ b/python/paddle/proto/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/PyDataProvider2.py b/python/paddle/trainer/PyDataProvider2.py index 0c577ec657..6e8cce1cce 100644 --- a/python/paddle/trainer/PyDataProvider2.py +++ b/python/paddle/trainer/PyDataProvider2.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/PyDataProviderWrapper.py b/python/paddle/trainer/PyDataProviderWrapper.py index 90b684a000..6af2507728 100644 --- a/python/paddle/trainer/PyDataProviderWrapper.py +++ b/python/paddle/trainer/PyDataProviderWrapper.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/__init__.py b/python/paddle/trainer/__init__.py index c90af2ee00..f662d68263 100644 --- a/python/paddle/trainer/__init__.py +++ b/python/paddle/trainer/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/config_parser.py b/python/paddle/trainer/config_parser.py index c6c0c9c151..fd7fb40822 100644 --- a/python/paddle/trainer/config_parser.py +++ b/python/paddle/trainer/config_parser.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/config_parser_extension.py b/python/paddle/trainer/config_parser_extension.py index ba4c79efdc..b9e0f3eb13 100644 --- a/python/paddle/trainer/config_parser_extension.py +++ b/python/paddle/trainer/config_parser_extension.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer/recurrent_units.py b/python/paddle/trainer/recurrent_units.py index a80ad13d1e..edca279dca 100644 --- a/python/paddle/trainer/recurrent_units.py +++ b/python/paddle/trainer/recurrent_units.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/__init__.py b/python/paddle/trainer_config_helpers/__init__.py index adebebba25..3ac1454934 100644 --- a/python/paddle/trainer_config_helpers/__init__.py +++ b/python/paddle/trainer_config_helpers/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/activations.py b/python/paddle/trainer_config_helpers/activations.py index eeed18a98a..06be3e4599 100644 --- a/python/paddle/trainer_config_helpers/activations.py +++ b/python/paddle/trainer_config_helpers/activations.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/attrs.py b/python/paddle/trainer_config_helpers/attrs.py index 54169f382f..59bb18bfca 100644 --- a/python/paddle/trainer_config_helpers/attrs.py +++ b/python/paddle/trainer_config_helpers/attrs.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/data_sources.py b/python/paddle/trainer_config_helpers/data_sources.py index b41097953d..b6ecd42857 100644 --- a/python/paddle/trainer_config_helpers/data_sources.py +++ b/python/paddle/trainer_config_helpers/data_sources.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/default_decorators.py b/python/paddle/trainer_config_helpers/default_decorators.py index c01050e338..1caad19349 100644 --- a/python/paddle/trainer_config_helpers/default_decorators.py +++ b/python/paddle/trainer_config_helpers/default_decorators.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/evaluators.py b/python/paddle/trainer_config_helpers/evaluators.py index dc6a36392f..0ee116d8c4 100644 --- a/python/paddle/trainer_config_helpers/evaluators.py +++ b/python/paddle/trainer_config_helpers/evaluators.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/layers.py b/python/paddle/trainer_config_helpers/layers.py index 4541b6fd8d..8dd6b7b7d2 100644 --- a/python/paddle/trainer_config_helpers/layers.py +++ b/python/paddle/trainer_config_helpers/layers.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/math.py b/python/paddle/trainer_config_helpers/math.py index 30a9b1c4e8..2d9e36f2b0 100644 --- a/python/paddle/trainer_config_helpers/math.py +++ b/python/paddle/trainer_config_helpers/math.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/networks.py b/python/paddle/trainer_config_helpers/networks.py index ff6d2e1cff..375bea34e8 100644 --- a/python/paddle/trainer_config_helpers/networks.py +++ b/python/paddle/trainer_config_helpers/networks.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/optimizers.py b/python/paddle/trainer_config_helpers/optimizers.py index 501fc3211b..d95b2cfe46 100644 --- a/python/paddle/trainer_config_helpers/optimizers.py +++ b/python/paddle/trainer_config_helpers/optimizers.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/poolings.py b/python/paddle/trainer_config_helpers/poolings.py index 6f13a76f25..0c38a8dce5 100644 --- a/python/paddle/trainer_config_helpers/poolings.py +++ b/python/paddle/trainer_config_helpers/poolings.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/tests/ProtobufEqualMain.cpp b/python/paddle/trainer_config_helpers/tests/ProtobufEqualMain.cpp index 06f7de9306..fc53422afd 100644 --- a/python/paddle/trainer_config_helpers/tests/ProtobufEqualMain.cpp +++ b/python/paddle/trainer_config_helpers/tests/ProtobufEqualMain.cpp @@ -1,4 +1,4 @@ -/* Copyright (c) 2016 Baidu, Inc. All Rights Reserve. +/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/tests/layers_test.py b/python/paddle/trainer_config_helpers/tests/layers_test.py index 3b55667354..05902ea293 100644 --- a/python/paddle/trainer_config_helpers/tests/layers_test.py +++ b/python/paddle/trainer_config_helpers/tests/layers_test.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/tests/layers_test_config.py b/python/paddle/trainer_config_helpers/tests/layers_test_config.py index 44d134d1f7..ae275735aa 100644 --- a/python/paddle/trainer_config_helpers/tests/layers_test_config.py +++ b/python/paddle/trainer_config_helpers/tests/layers_test_config.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/trainer_config_helpers/utils.py b/python/paddle/trainer_config_helpers/utils.py index c0235b28cd..fe6e9cd53c 100644 --- a/python/paddle/trainer_config_helpers/utils.py +++ b/python/paddle/trainer_config_helpers/utils.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/__init__.py b/python/paddle/utils/__init__.py index 3e93f41c2e..15595d2085 100644 --- a/python/paddle/utils/__init__.py +++ b/python/paddle/utils/__init__.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/dump_config.py b/python/paddle/utils/dump_config.py index c5ce5c8d9a..73bf349c46 100644 --- a/python/paddle/utils/dump_config.py +++ b/python/paddle/utils/dump_config.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/image_util.py b/python/paddle/utils/image_util.py index e6c6b04de0..d3d79b1440 100644 --- a/python/paddle/utils/image_util.py +++ b/python/paddle/utils/image_util.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/make_model_diagram.py b/python/paddle/utils/make_model_diagram.py index 29e271717d..1370ea83a4 100644 --- a/python/paddle/utils/make_model_diagram.py +++ b/python/paddle/utils/make_model_diagram.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/plotcurve.py b/python/paddle/utils/plotcurve.py index 7bc7c5f8d2..27bd8157d3 100644 --- a/python/paddle/utils/plotcurve.py +++ b/python/paddle/utils/plotcurve.py @@ -1,5 +1,5 @@ #!/usr/bin/python -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/predefined_net.py b/python/paddle/utils/predefined_net.py index e9033432ed..fa05f981f2 100644 --- a/python/paddle/utils/predefined_net.py +++ b/python/paddle/utils/predefined_net.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/preprocess_img.py b/python/paddle/utils/preprocess_img.py index f3c609e4cd..975f1e9ede 100644 --- a/python/paddle/utils/preprocess_img.py +++ b/python/paddle/utils/preprocess_img.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/preprocess_util.py b/python/paddle/utils/preprocess_util.py index e5067a80ea..1d17a48824 100644 --- a/python/paddle/utils/preprocess_util.py +++ b/python/paddle/utils/preprocess_util.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/show_pb.py b/python/paddle/utils/show_pb.py index 3b371727b8..20614826d1 100644 --- a/python/paddle/utils/show_pb.py +++ b/python/paddle/utils/show_pb.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/python/paddle/utils/torch2paddle.py b/python/paddle/utils/torch2paddle.py index 958f55dbc4..91490111a1 100644 --- a/python/paddle/utils/torch2paddle.py +++ b/python/paddle/utils/torch2paddle.py @@ -1,4 +1,4 @@ -# Copyright (c) 2016 Baidu, Inc. All Rights Reserved +# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. From 89f0072a38caec86ffe03e4437bc84434d7c34b7 Mon Sep 17 00:00:00 2001 From: Luo Tao Date: Fri, 9 Dec 2016 12:02:57 +0800 Subject: [PATCH 36/60] add file extensions in doc/ --- doc/about/{index.rst => index_en.rst} | 0 .../data_provider/{index.rst => index_en.rst} | 8 ------- ...taprovider2.rst => pydataprovider2_en.rst} | 0 doc/api/{index.rst => index_en.rst} | 7 +++--- .../{basic_usage.rst => index_en.rst} | 0 ...from_source.md => build_from_source_en.md} | 0 ...cker_install.rst => docker_install_en.rst} | 0 .../{index.rst => index_en.rst} | 8 +++---- ...untu_install.rst => ubuntu_install_en.rst} | 0 doc/getstarted/index.rst | 8 ------- doc/getstarted/index_en.rst | 8 +++++++ .../{cluster_train.md => cluster_train_en.md} | 0 .../{arguments.md => arguments_en.md} | 0 ...roduction.md => detail_introduction_en.md} | 0 doc/howto/cmd_parameter/index.md | 5 ---- doc/howto/cmd_parameter/index_en.md | 5 ++++ .../{use_case.md => use_case_en.md} | 0 ...o_paddle.md => contribute_to_paddle_en.md} | 0 .../deep_model/{index.rst => index_en.rst} | 2 +- .../deep_model/rnn/{rnn.rst => rnn_en.rst} | 0 doc/howto/{index.rst => index_en.rst} | 12 +++++----- .../new_layer/{index.rst => index_en.rst} | 0 ...gpu_profiling.rst => gpu_profiling_en.rst} | 0 .../optimization/{index.rst => index_en.rst} | 2 +- doc/index.rst | 11 ++++----- .../embedding_model/{index.md => index_en.md} | 0 doc/tutorials/image_classification/index.rst | 10 -------- .../{image_classification.md => index_en.md} | 0 .../{resnet_model.md => resnet_model_en.md} | 0 doc/tutorials/index.md | 21 ----------------- doc/tutorials/index_en.md | 23 +++++++++++++++++++ .../rec/{ml_dataset.md => ml_dataset_en.md} | 0 ...ml_regression.rst => ml_regression_en.rst} | 0 .../semantic_role_labeling/index.rst | 7 ------ ...{semantic_role_labeling.md => index_en.md} | 0 doc/tutorials/sentiment_analysis/index.rst | 9 -------- .../{sentiment_analysis.md => index_en.md} | 0 doc/tutorials/text_generation/index.rst | 9 -------- .../{text_generation.md => index_en.md} | 0 39 files changed, 55 insertions(+), 100 deletions(-) rename doc/about/{index.rst => index_en.rst} (100%) rename doc/api/data_provider/{index.rst => index_en.rst} (89%) rename doc/api/data_provider/{pydataprovider2.rst => pydataprovider2_en.rst} (100%) rename doc/api/{index.rst => index_en.rst} (80%) rename doc/getstarted/basic_usage/{basic_usage.rst => index_en.rst} (100%) rename doc/getstarted/build_and_install/{build_from_source.md => build_from_source_en.md} (100%) rename doc/getstarted/build_and_install/{docker_install.rst => docker_install_en.rst} (100%) rename doc/getstarted/build_and_install/{index.rst => index_en.rst} (79%) rename doc/getstarted/build_and_install/{ubuntu_install.rst => ubuntu_install_en.rst} (100%) delete mode 100644 doc/getstarted/index.rst create mode 100644 doc/getstarted/index_en.rst rename doc/howto/cluster/{cluster_train.md => cluster_train_en.md} (100%) rename doc/howto/cmd_parameter/{arguments.md => arguments_en.md} (100%) rename doc/howto/cmd_parameter/{detail_introduction.md => detail_introduction_en.md} (100%) delete mode 100644 doc/howto/cmd_parameter/index.md create mode 100644 doc/howto/cmd_parameter/index_en.md rename doc/howto/cmd_parameter/{use_case.md => use_case_en.md} (100%) rename doc/howto/{contribute_to_paddle.md => contribute_to_paddle_en.md} (100%) rename doc/howto/deep_model/{index.rst => index_en.rst} (83%) rename doc/howto/deep_model/rnn/{rnn.rst => rnn_en.rst} (100%) rename doc/howto/{index.rst => index_en.rst} (51%) rename doc/howto/new_layer/{index.rst => index_en.rst} (100%) rename doc/howto/optimization/{gpu_profiling.rst => gpu_profiling_en.rst} (100%) rename doc/howto/optimization/{index.rst => index_en.rst} (78%) rename doc/tutorials/embedding_model/{index.md => index_en.md} (100%) delete mode 100644 doc/tutorials/image_classification/index.rst rename doc/tutorials/image_classification/{image_classification.md => index_en.md} (100%) rename doc/tutorials/imagenet_model/{resnet_model.md => resnet_model_en.md} (100%) delete mode 100644 doc/tutorials/index.md create mode 100644 doc/tutorials/index_en.md rename doc/tutorials/rec/{ml_dataset.md => ml_dataset_en.md} (100%) rename doc/tutorials/rec/{ml_regression.rst => ml_regression_en.rst} (100%) delete mode 100644 doc/tutorials/semantic_role_labeling/index.rst rename doc/tutorials/semantic_role_labeling/{semantic_role_labeling.md => index_en.md} (100%) delete mode 100644 doc/tutorials/sentiment_analysis/index.rst rename doc/tutorials/sentiment_analysis/{sentiment_analysis.md => index_en.md} (100%) delete mode 100644 doc/tutorials/text_generation/index.rst rename doc/tutorials/text_generation/{text_generation.md => index_en.md} (100%) diff --git a/doc/about/index.rst b/doc/about/index_en.rst similarity index 100% rename from doc/about/index.rst rename to doc/about/index_en.rst diff --git a/doc/api/data_provider/index.rst b/doc/api/data_provider/index_en.rst similarity index 89% rename from doc/api/data_provider/index.rst rename to doc/api/data_provider/index_en.rst index 5e7a49d632..96efbb1da9 100644 --- a/doc/api/data_provider/index.rst +++ b/doc/api/data_provider/index_en.rst @@ -32,11 +32,3 @@ Each line of train.list and test.list is an absolute or relative path (relative to the PaddePaddle program runtime) of data file. Fascinatingly more, each line can also be a HDFS file path or a SQL connection string. As long as the user assures how to access each file in DataProvider. - -Please refer to the following articles for more information about the detail -usages of DataProvider and how to implement a new DataProvider, - -.. toctree:: - - pydataprovider2.rst - write_new_dataprovider.rst diff --git a/doc/api/data_provider/pydataprovider2.rst b/doc/api/data_provider/pydataprovider2_en.rst similarity index 100% rename from doc/api/data_provider/pydataprovider2.rst rename to doc/api/data_provider/pydataprovider2_en.rst diff --git a/doc/api/index.rst b/doc/api/index_en.rst similarity index 80% rename from doc/api/index.rst rename to doc/api/index_en.rst index ccee7a0f1f..9930f93e10 100644 --- a/doc/api/index.rst +++ b/doc/api/index_en.rst @@ -7,8 +7,8 @@ DataProvider API .. toctree:: :maxdepth: 1 - data_provider/index.rst - data_provider/pydataprovider2.rst + data_provider/index_en.rst + data_provider/pydataprovider2_en.rst Model Config API ---------------- @@ -16,7 +16,6 @@ Model Config API .. toctree:: :maxdepth: 1 - trainer_config_helpers/index.rst trainer_config_helpers/optimizers.rst trainer_config_helpers/data_sources.rst trainer_config_helpers/layers.rst @@ -33,4 +32,4 @@ Applications API .. toctree:: :maxdepth: 1 - predict/swig_py_paddle_en.rst \ No newline at end of file + predict/swig_py_paddle_en.rst diff --git a/doc/getstarted/basic_usage/basic_usage.rst b/doc/getstarted/basic_usage/index_en.rst similarity index 100% rename from doc/getstarted/basic_usage/basic_usage.rst rename to doc/getstarted/basic_usage/index_en.rst diff --git a/doc/getstarted/build_and_install/build_from_source.md b/doc/getstarted/build_and_install/build_from_source_en.md similarity index 100% rename from doc/getstarted/build_and_install/build_from_source.md rename to doc/getstarted/build_and_install/build_from_source_en.md diff --git a/doc/getstarted/build_and_install/docker_install.rst b/doc/getstarted/build_and_install/docker_install_en.rst similarity index 100% rename from doc/getstarted/build_and_install/docker_install.rst rename to doc/getstarted/build_and_install/docker_install_en.rst diff --git a/doc/getstarted/build_and_install/index.rst b/doc/getstarted/build_and_install/index_en.rst similarity index 79% rename from doc/getstarted/build_and_install/index.rst rename to doc/getstarted/build_and_install/index_en.rst index 6187be9d72..1bfd4f75c0 100644 --- a/doc/getstarted/build_and_install/index.rst +++ b/doc/getstarted/build_and_install/index_en.rst @@ -6,10 +6,9 @@ Install PaddlePaddle .. toctree:: :maxdepth: 1 - :glob: - docker_install.rst - ubuntu_install.rst + docker_install_en.rst + ubuntu_install_en.rst Build from Source ----------------- @@ -20,6 +19,5 @@ Build from Source .. toctree:: :maxdepth: 1 - :glob: - build_from_source.md \ No newline at end of file + build_from_source_en.md diff --git a/doc/getstarted/build_and_install/ubuntu_install.rst b/doc/getstarted/build_and_install/ubuntu_install_en.rst similarity index 100% rename from doc/getstarted/build_and_install/ubuntu_install.rst rename to doc/getstarted/build_and_install/ubuntu_install_en.rst diff --git a/doc/getstarted/index.rst b/doc/getstarted/index.rst deleted file mode 100644 index 5f2787066e..0000000000 --- a/doc/getstarted/index.rst +++ /dev/null @@ -1,8 +0,0 @@ -GET STARTED -============ - -.. toctree:: - :maxdepth: 2 - - build_and_install/index.rst - basic_usage/basic_usage.rst diff --git a/doc/getstarted/index_en.rst b/doc/getstarted/index_en.rst new file mode 100644 index 0000000000..55d95d8015 --- /dev/null +++ b/doc/getstarted/index_en.rst @@ -0,0 +1,8 @@ +GET STARTED +============ + +.. toctree:: + :maxdepth: 2 + + build_and_install/index_en.rst + basic_usage/index_en.rst diff --git a/doc/howto/cluster/cluster_train.md b/doc/howto/cluster/cluster_train_en.md similarity index 100% rename from doc/howto/cluster/cluster_train.md rename to doc/howto/cluster/cluster_train_en.md diff --git a/doc/howto/cmd_parameter/arguments.md b/doc/howto/cmd_parameter/arguments_en.md similarity index 100% rename from doc/howto/cmd_parameter/arguments.md rename to doc/howto/cmd_parameter/arguments_en.md diff --git a/doc/howto/cmd_parameter/detail_introduction.md b/doc/howto/cmd_parameter/detail_introduction_en.md similarity index 100% rename from doc/howto/cmd_parameter/detail_introduction.md rename to doc/howto/cmd_parameter/detail_introduction_en.md diff --git a/doc/howto/cmd_parameter/index.md b/doc/howto/cmd_parameter/index.md deleted file mode 100644 index 48cf835de1..0000000000 --- a/doc/howto/cmd_parameter/index.md +++ /dev/null @@ -1,5 +0,0 @@ -# How to Set Command-line Parameters - -* [Use Case](use_case.md) -* [Arguments](arguments.md) -* [Detailed Descriptions](detail_introduction.md) diff --git a/doc/howto/cmd_parameter/index_en.md b/doc/howto/cmd_parameter/index_en.md new file mode 100644 index 0000000000..bd16affdd8 --- /dev/null +++ b/doc/howto/cmd_parameter/index_en.md @@ -0,0 +1,5 @@ +# How to Set Command-line Parameters + +* [Use Case](use_case_en.md) +* [Arguments](arguments_en.md) +* [Detailed Descriptions](detail_introduction_en.md) diff --git a/doc/howto/cmd_parameter/use_case.md b/doc/howto/cmd_parameter/use_case_en.md similarity index 100% rename from doc/howto/cmd_parameter/use_case.md rename to doc/howto/cmd_parameter/use_case_en.md diff --git a/doc/howto/contribute_to_paddle.md b/doc/howto/contribute_to_paddle_en.md similarity index 100% rename from doc/howto/contribute_to_paddle.md rename to doc/howto/contribute_to_paddle_en.md diff --git a/doc/howto/deep_model/index.rst b/doc/howto/deep_model/index_en.rst similarity index 83% rename from doc/howto/deep_model/index.rst rename to doc/howto/deep_model/index_en.rst index 06ef443f62..00a45641e6 100644 --- a/doc/howto/deep_model/index.rst +++ b/doc/howto/deep_model/index_en.rst @@ -4,4 +4,4 @@ How to Configure Deep Models .. toctree:: :maxdepth: 1 - rnn/rnn.rst + rnn/rnn_en.rst diff --git a/doc/howto/deep_model/rnn/rnn.rst b/doc/howto/deep_model/rnn/rnn_en.rst similarity index 100% rename from doc/howto/deep_model/rnn/rnn.rst rename to doc/howto/deep_model/rnn/rnn_en.rst diff --git a/doc/howto/index.rst b/doc/howto/index_en.rst similarity index 51% rename from doc/howto/index.rst rename to doc/howto/index_en.rst index 41877a64a5..bd64c5b1fb 100644 --- a/doc/howto/index.rst +++ b/doc/howto/index_en.rst @@ -7,9 +7,9 @@ Usage .. toctree:: :maxdepth: 1 - cmd_parameter/index.md - deep_model/index.rst - cluster/cluster_train.md + cmd_parameter/index_en.md + deep_model/index_en.rst + cluster/cluster_train_en.md Development ------------ @@ -17,8 +17,8 @@ Development .. toctree:: :maxdepth: 1 - new_layer/index.rst - contribute_to_paddle.md + new_layer/index_en.rst + contribute_to_paddle_en.md Optimization ------------- @@ -26,4 +26,4 @@ Optimization .. toctree:: :maxdepth: 1 - optimization/index.rst + optimization/index_en.rst diff --git a/doc/howto/new_layer/index.rst b/doc/howto/new_layer/index_en.rst similarity index 100% rename from doc/howto/new_layer/index.rst rename to doc/howto/new_layer/index_en.rst diff --git a/doc/howto/optimization/gpu_profiling.rst b/doc/howto/optimization/gpu_profiling_en.rst similarity index 100% rename from doc/howto/optimization/gpu_profiling.rst rename to doc/howto/optimization/gpu_profiling_en.rst diff --git a/doc/howto/optimization/index.rst b/doc/howto/optimization/index_en.rst similarity index 78% rename from doc/howto/optimization/index.rst rename to doc/howto/optimization/index_en.rst index e2822a0098..1e2f16b5da 100644 --- a/doc/howto/optimization/index.rst +++ b/doc/howto/optimization/index_en.rst @@ -4,4 +4,4 @@ How to Tune GPU Performance .. toctree:: :maxdepth: 3 - gpu_profiling.rst + gpu_profiling_en.rst diff --git a/doc/index.rst b/doc/index.rst index 3555da1dfc..c107239438 100644 --- a/doc/index.rst +++ b/doc/index.rst @@ -4,9 +4,8 @@ PaddlePaddle Documentation .. toctree:: :maxdepth: 1 - getstarted/index.rst - tutorials/index.md - howto/index.rst - api/index.rst - about/index.rst - \ No newline at end of file + getstarted/index_en.rst + tutorials/index_en.md + howto/index_en.rst + api/index_en.rst + about/index_en.rst diff --git a/doc/tutorials/embedding_model/index.md b/doc/tutorials/embedding_model/index_en.md similarity index 100% rename from doc/tutorials/embedding_model/index.md rename to doc/tutorials/embedding_model/index_en.md diff --git a/doc/tutorials/image_classification/index.rst b/doc/tutorials/image_classification/index.rst deleted file mode 100644 index 1ea68f1416..0000000000 --- a/doc/tutorials/image_classification/index.rst +++ /dev/null @@ -1,10 +0,0 @@ -Image Classification Tutorial -============================= - -.. toctree:: - :maxdepth: 3 - :glob: - - Training Locally - cluster_train/internal/cluster_train.md - cluster_train/opensource/cluster_train.md diff --git a/doc/tutorials/image_classification/image_classification.md b/doc/tutorials/image_classification/index_en.md similarity index 100% rename from doc/tutorials/image_classification/image_classification.md rename to doc/tutorials/image_classification/index_en.md diff --git a/doc/tutorials/imagenet_model/resnet_model.md b/doc/tutorials/imagenet_model/resnet_model_en.md similarity index 100% rename from doc/tutorials/imagenet_model/resnet_model.md rename to doc/tutorials/imagenet_model/resnet_model_en.md diff --git a/doc/tutorials/index.md b/doc/tutorials/index.md deleted file mode 100644 index ebf5397391..0000000000 --- a/doc/tutorials/index.md +++ /dev/null @@ -1,21 +0,0 @@ -# TUTORIALS -There are serveral examples and demos here. - -## Image - -* [Image Classification](image_classification/index.rst) - -## NLP - -* [Sentiment Analysis](sentiment_analysis/index.rst) -* [Text Generation](text_generation/index.rst) -* [Semantic Role Labeling](semantic_role_labeling/index.rst) - -## Recommendation - -* [MovieLens Dataset](rec/ml_dataset.md) -* [MovieLens Regression](rec/ml_regression.rst) - -## Model Zoo -* [ImageNet: ResNet](imagenet_model/resnet_model.md) -* [Embedding: Chinese Word](embedding_model/index.md) diff --git a/doc/tutorials/index_en.md b/doc/tutorials/index_en.md new file mode 100644 index 0000000000..97de356665 --- /dev/null +++ b/doc/tutorials/index_en.md @@ -0,0 +1,23 @@ +# TUTORIALS +There are serveral examples and demos here. + +## [Quick Start](quick_start/index_en.md) + +## Image + +* [Image Classification](image_classification/index_en.md) + +## NLP + +* [Sentiment Analysis](sentiment_analysis/index_en.md) +* [Text Generation](text_generation/index_en.md) +* [Semantic Role Labeling](semantic_role_labeling/index_en.md) + +## Recommendation + +* [MovieLens Dataset](rec/ml_dataset_en.md) +* [MovieLens Regression](rec/ml_regression_en.rst) + +## Model Zoo +* [ImageNet: ResNet](imagenet_model/resnet_model_en.md) +* [Embedding: Chinese Word](embedding_model/index_en.md) diff --git a/doc/tutorials/rec/ml_dataset.md b/doc/tutorials/rec/ml_dataset_en.md similarity index 100% rename from doc/tutorials/rec/ml_dataset.md rename to doc/tutorials/rec/ml_dataset_en.md diff --git a/doc/tutorials/rec/ml_regression.rst b/doc/tutorials/rec/ml_regression_en.rst similarity index 100% rename from doc/tutorials/rec/ml_regression.rst rename to doc/tutorials/rec/ml_regression_en.rst diff --git a/doc/tutorials/semantic_role_labeling/index.rst b/doc/tutorials/semantic_role_labeling/index.rst deleted file mode 100644 index ff3035059b..0000000000 --- a/doc/tutorials/semantic_role_labeling/index.rst +++ /dev/null @@ -1,7 +0,0 @@ -Semantic Role Labeling Tutorial -=============================== - -.. toctree:: - :maxdepth: 3 - - semantic_role_labeling.md diff --git a/doc/tutorials/semantic_role_labeling/semantic_role_labeling.md b/doc/tutorials/semantic_role_labeling/index_en.md similarity index 100% rename from doc/tutorials/semantic_role_labeling/semantic_role_labeling.md rename to doc/tutorials/semantic_role_labeling/index_en.md diff --git a/doc/tutorials/sentiment_analysis/index.rst b/doc/tutorials/sentiment_analysis/index.rst deleted file mode 100644 index 9ee6d3a177..0000000000 --- a/doc/tutorials/sentiment_analysis/index.rst +++ /dev/null @@ -1,9 +0,0 @@ -Sentiment Analasis Tutorial -=========================== - -.. toctree:: - :maxdepth: 3 - :glob: - - Training Locally - internal/cluster_train.md diff --git a/doc/tutorials/sentiment_analysis/sentiment_analysis.md b/doc/tutorials/sentiment_analysis/index_en.md similarity index 100% rename from doc/tutorials/sentiment_analysis/sentiment_analysis.md rename to doc/tutorials/sentiment_analysis/index_en.md diff --git a/doc/tutorials/text_generation/index.rst b/doc/tutorials/text_generation/index.rst deleted file mode 100644 index 82da552419..0000000000 --- a/doc/tutorials/text_generation/index.rst +++ /dev/null @@ -1,9 +0,0 @@ -Text Generation Tutorial -======================== - -.. toctree:: - :maxdepth: 3 - :glob: - - Training Locally - internal/cluster_train.md diff --git a/doc/tutorials/text_generation/text_generation.md b/doc/tutorials/text_generation/index_en.md similarity index 100% rename from doc/tutorials/text_generation/text_generation.md rename to doc/tutorials/text_generation/index_en.md From 3d2896493b90283b6b5b8e49869d90f656f39783 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Fri, 9 Dec 2016 14:00:04 +0800 Subject: [PATCH 37/60] follow comments --- demo/sentiment/predict.py | 62 ++++++++----------- .../sentiment_analysis/sentiment_analysis.md | 29 ++++----- .../sentiment_analysis/sentiment_analysis.md | 29 ++++----- 3 files changed, 57 insertions(+), 63 deletions(-) diff --git a/demo/sentiment/predict.py b/demo/sentiment/predict.py index e01dc6d228..3920a1bade 100755 --- a/demo/sentiment/predict.py +++ b/demo/sentiment/predict.py @@ -66,42 +66,27 @@ class SentimentPrediction(): for v in open(label_file, 'r'): self.label[int(v.split('\t')[1])] = v.split('\t')[0] - def get_data(self, data): + def get_index(self, data): """ - Get input data of paddle format. + transform word into integer index according to the dictionary. """ - for line in data: - words = line.strip().split() - word_slot = [ - self.word_dict[w] for w in words if w in self.word_dict - ] - if not word_slot: - print "all words are not in dictionary: %s", line - continue - yield [word_slot] - - def predict(self, batch_size): - - def batch_predict(batch_data): - input = self.converter(self.get_data(batch_data)) - output = self.network.forwardTest(input) - prob = output[0]["value"] - labs = np.argsort(-prob) - for idx, lab in enumerate(labs): - if self.label is None: - print("predicting label is %d" % (lab[0])) - else: - print("predicting label is %s" % - (self.label[lab[0]])) - - batch = [] - for line in sys.stdin: - batch.append(line) - if len(batch) == batch_size: - batch_predict(batch) - batch=[] - if len(batch) > 0: - batch_predict(batch) + words = data.strip().split() + word_slot = [ + self.word_dict[w] for w in words if w in self.word_dict + ] + return word_slot + + def batch_predict(self, data_batch): + input = self.converter(data_batch) + output = self.network.forwardTest(input) + prob = output[0]["value"] + labs = np.argsort(-prob) + for idx, lab in enumerate(labs): + if self.label is None: + print("predicting label is %d" % (lab[0])) + else: + print("predicting label is %s" % + (self.label[lab[0]])) def option_parser(): usage = "python predict.py -n config -w model_dir -d dictionary -i input_file " @@ -152,8 +137,15 @@ def main(): label = options.label swig_paddle.initPaddle("--use_gpu=0") predict = SentimentPrediction(train_conf, dict_file, model_path, label) - predict.predict(batch_size) + batch = [] + for line in sys.stdin: + batch.append([predict.get_index(line)]) + if len(batch) == batch_size: + predict.batch_predict(batch) + batch=[] + if len(batch) > 0: + predict.batch_predict(batch) if __name__ == '__main__': main() diff --git a/doc/tutorials/sentiment_analysis/sentiment_analysis.md b/doc/tutorials/sentiment_analysis/sentiment_analysis.md index c53952c544..bb7681db44 100644 --- a/doc/tutorials/sentiment_analysis/sentiment_analysis.md +++ b/doc/tutorials/sentiment_analysis/sentiment_analysis.md @@ -293,20 +293,21 @@ predict.sh: model=model_output/pass-00002/ config=trainer_config.py label=data/pre-imdb/labels.list -python predict.py \ - -n $config\ - -w $model \ - -b $label \ - -d data/pre-imdb/dict.txt \ - -i data/aclImdb/test/pos/10007_10.txt -``` - -* `predict.py`: predicting interface. -* -n $config : set network configure. -* -w $model: set model path. -* -b $label: set dictionary about corresponding relation between integer label and string label. -* -d data/pre-imdb/dict.txt: set dictionary. -* -i data/aclImdb/test/pos/10014_7.txt: set one example file to predict. +cat ./data/aclImdb/test/pos/10007_10.txt | python predict.py \ + --tconf=$config\ + --model=$model \ + --label=$label \ + --dict=./data/pre-imdb/dict.txt \ + --batch_size=1 +``` + +* `cat ./data/aclImdb/test/pos/10007_10.txt` : the input sample. +* `predict.py` : predicting interface. +* `--tconf=$config` : set network configure. +* ` --model=$model` : set model path. +* `--label=$label` : set dictionary about corresponding relation between integer label and string label. +* `--dict=data/pre-imdb/dict.txt` : set dictionary. +* `--batch_size=1` : set batch size. Note you should make sure the default model path `model_output/pass-00002` exists or change the model path. diff --git a/doc_cn/demo/sentiment_analysis/sentiment_analysis.md b/doc_cn/demo/sentiment_analysis/sentiment_analysis.md index b70f2d5967..ba307e97e3 100644 --- a/doc_cn/demo/sentiment_analysis/sentiment_analysis.md +++ b/doc_cn/demo/sentiment_analysis/sentiment_analysis.md @@ -291,20 +291,21 @@ predict.sh: model=model_output/pass-00002/ config=trainer_config.py label=data/pre-imdb/labels.list -python predict.py \ - -n $config\ - -w $model \ - -b $label \ - -d data/pre-imdb/dict.txt \ - -i data/aclImdb/test/pos/10007_10.txt -``` - -* `predict.py`: 预测接口脚本。 -* -n $config : 设置网络配置。 -* -w $model: 设置模型路径。 -* -b $label: 设置标签类别字典，这个字典是整数标签和字符串标签的一个对应。 -* -d data/pre-imdb/dict.txt: 设置字典文件。 -* -i data/aclImdb/test/pos/10014_7.txt: 设置一个要预测的示例文件。 +cat ./data/aclImdb/test/pos/10007_10.txt | python predict.py \ + --tconf=$config\ + --model=$model \ + --label=$label \ + --dict=./data/pre-imdb/dict.txt \ + --batch_size=1 +``` + +* `cat ./data/aclImdb/test/pos/10007_10.txt` : 输入预测样本。 +* `predict.py` : 预测接口脚本。 +* `--tconf=$config` : 设置网络配置。 +* `--model=$model` : 设置模型路径。 +* `--label=$label` : 设置标签类别字典，这个字典是整数标签和字符串标签的一个对应。 +* `--dict=data/pre-imdb/dict.txt` : 设置字典文件。 +* `--batch_size=1` : 设置batch size。 注意应该确保默认模型路径`model_output / pass-00002`存在或更改为其它模型路径。 From 0c96b26c3ce3e7f1a3e6c9501d9675874d846564 Mon Sep 17 00:00:00 2001 From: Luo Tao Date: Fri, 9 Dec 2016 17:01:59 +0800 Subject: [PATCH 38/60] remove breathe and doc/howto/source directory --- .travis.yml | 2 +- cmake/check_packages.cmake | 1 - .../build_and_install/build_from_source_en.md | 2 +- doc/howto/source/api.rst | 5 - doc/howto/source/cuda/index.rst | 9 - doc/howto/source/cuda/matrix.rst | 59 -- doc/howto/source/cuda/nn.rst | 39 -- doc/howto/source/cuda/utils.rst | 37 -- doc/howto/source/gserver/activations.rst | 5 - doc/howto/source/gserver/dataproviders.rst | 87 --- doc/howto/source/gserver/evaluators.rst | 103 ---- doc/howto/source/gserver/gradientmachines.rst | 27 - doc/howto/source/gserver/index.rst | 12 - doc/howto/source/gserver/layers.rst | 566 ------------------ doc/howto/source/gserver/neworks.rst | 12 - doc/howto/source/index.rst | 14 - doc/howto/source/math/functions.rst | 10 - doc/howto/source/math/index.rst | 10 - doc/howto/source/math/matrix.rst | 76 --- doc/howto/source/math/utils.rst | 18 - doc/howto/source/math/vector.rst | 37 -- doc/howto/source/parameter/index.rst | 9 - doc/howto/source/parameter/optimizer.rst | 22 - doc/howto/source/parameter/parameter.rst | 12 - doc/howto/source/parameter/updater.rst | 14 - doc/howto/source/pserver/client.rst | 12 - doc/howto/source/pserver/index.rst | 10 - doc/howto/source/pserver/network.rst | 27 - doc/howto/source/pserver/server.rst | 12 - doc/howto/source/trainer.rst | 32 - doc/howto/source/utils/customStackTrace.rst | 4 - doc/howto/source/utils/enum.rst | 3 - doc/howto/source/utils/index.rst | 11 - doc/howto/source/utils/lock.rst | 32 - doc/howto/source/utils/queue.rst | 12 - doc/howto/source/utils/thread.rst | 27 - paddle/scripts/docker/Dockerfile | 2 +- paddle/scripts/docker/Dockerfile.gpu | 2 +- paddle/scripts/tools/build_docs/Dockerfile | 2 +- 39 files changed, 5 insertions(+), 1371 deletions(-) delete mode 100644 doc/howto/source/api.rst delete mode 100644 doc/howto/source/cuda/index.rst delete mode 100644 doc/howto/source/cuda/matrix.rst delete mode 100644 doc/howto/source/cuda/nn.rst delete mode 100644 doc/howto/source/cuda/utils.rst delete mode 100644 doc/howto/source/gserver/activations.rst delete mode 100644 doc/howto/source/gserver/dataproviders.rst delete mode 100644 doc/howto/source/gserver/evaluators.rst delete mode 100644 doc/howto/source/gserver/gradientmachines.rst delete mode 100644 doc/howto/source/gserver/index.rst delete mode 100644 doc/howto/source/gserver/layers.rst delete mode 100644 doc/howto/source/gserver/neworks.rst delete mode 100644 doc/howto/source/index.rst delete mode 100644 doc/howto/source/math/functions.rst delete mode 100644 doc/howto/source/math/index.rst delete mode 100644 doc/howto/source/math/matrix.rst delete mode 100644 doc/howto/source/math/utils.rst delete mode 100644 doc/howto/source/math/vector.rst delete mode 100644 doc/howto/source/parameter/index.rst delete mode 100644 doc/howto/source/parameter/optimizer.rst delete mode 100644 doc/howto/source/parameter/parameter.rst delete mode 100644 doc/howto/source/parameter/updater.rst delete mode 100644 doc/howto/source/pserver/client.rst delete mode 100644 doc/howto/source/pserver/index.rst delete mode 100644 doc/howto/source/pserver/network.rst delete mode 100644 doc/howto/source/pserver/server.rst delete mode 100644 doc/howto/source/trainer.rst delete mode 100644 doc/howto/source/utils/customStackTrace.rst delete mode 100644 doc/howto/source/utils/enum.rst delete mode 100644 doc/howto/source/utils/index.rst delete mode 100644 doc/howto/source/utils/lock.rst delete mode 100644 doc/howto/source/utils/queue.rst delete mode 100644 doc/howto/source/utils/thread.rst diff --git a/.travis.yml b/.travis.yml index 6215060e33..cf0cca1134 100644 --- a/.travis.yml +++ b/.travis.yml @@ -50,7 +50,7 @@ before_install: fi - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then sudo paddle/scripts/travis/before_install.linux.sh; fi - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then paddle/scripts/travis/before_install.osx.sh; fi - - pip install wheel protobuf sphinx breathe recommonmark virtualenv numpy sphinx_rtd_theme + - pip install wheel protobuf sphinx recommonmark virtualenv numpy sphinx_rtd_theme script: - paddle/scripts/travis/main.sh notifications: diff --git a/cmake/check_packages.cmake b/cmake/check_packages.cmake index 3bc0c1fd18..0688745541 100644 --- a/cmake/check_packages.cmake +++ b/cmake/check_packages.cmake @@ -30,7 +30,6 @@ if(WITH_DOC) find_package(Sphinx REQUIRED) find_package(Doxygen REQUIRED) find_python_module(recommonmark REQUIRED) - find_python_module(breathe REQUIRED) endif() if(WITH_SWIG_PY) diff --git a/doc/getstarted/build_and_install/build_from_source_en.md b/doc/getstarted/build_and_install/build_from_source_en.md index 150d7fc437..3771d316a1 100644 --- a/doc/getstarted/build_and_install/build_from_source_en.md +++ b/doc/getstarted/build_and_install/build_from_source_en.md @@ -79,7 +79,7 @@ As a simple example, consider the following: ```bash pip install 'sphinx>=1.4.0' - pip install sphinx_rtd_theme breathe recommonmark + pip install sphinx_rtd_theme recommonmark # install doxygen on Ubuntu sudo apt-get install doxygen diff --git a/doc/howto/source/api.rst b/doc/howto/source/api.rst deleted file mode 100644 index 30396c26b6..0000000000 --- a/doc/howto/source/api.rst +++ /dev/null @@ -1,5 +0,0 @@ -API -=== - -.. doxygenfile:: paddle/api/PaddleAPI.h -.. doxygenfile:: paddle/api/Internal.h diff --git a/doc/howto/source/cuda/index.rst b/doc/howto/source/cuda/index.rst deleted file mode 100644 index b0fed2e7f7..0000000000 --- a/doc/howto/source/cuda/index.rst +++ /dev/null @@ -1,9 +0,0 @@ -CUDA -==== - -.. toctree:: - :maxdepth: 2 - - matrix.rst - nn.rst - utils.rst diff --git a/doc/howto/source/cuda/matrix.rst b/doc/howto/source/cuda/matrix.rst deleted file mode 100644 index b7699c83ed..0000000000 --- a/doc/howto/source/cuda/matrix.rst +++ /dev/null @@ -1,59 +0,0 @@ -Matrix -====== - -Base ----- - -hl_matrix.h -``````````` -.. doxygenfile:: paddle/cuda/include/hl_matrix.h - -hl_matrix_base.h -```````````````` -.. doxygenfile:: paddle/cuda/include/hl_matrix_base.cuh - -hl_matrix_apply.cuh -``````````````````` -.. doxygenfile:: paddle/cuda/include/hl_matrix_apply.cuh - -hl_matrix_ops.cuh -````````````````` -.. doxygenfile:: paddle/cuda/include/hl_matrix_ops.cuh - -hl_matrix_type.cuh -`````````````````` -.. doxygenfile:: paddle/cuda/include/hl_matrix_type.cuh - -hl_sse_matrix_kernel.cuh -```````````````````````` -.. doxygenfile:: paddle/cuda/include/hl_sse_matrix_kernel.cuh - -Matrix Function ---------------- - -hl_batch_transpose.h -```````````````````` -.. doxygenfile:: paddle/cuda/include/hl_batch_transpose.h - -hl_aggregate.h -`````````````` -.. doxygenfile:: paddle/cuda/include/hl_aggregate.h - -hl_top_k.h -`````````` -.. doxygenfile:: paddle/cuda/include/hl_top_k.h - -hl_table_apply.h -```````````````` -.. doxygenfile:: paddle/cuda/include/hl_table_apply.h - -Sparse Matrix -------------- - -hl_sparse.h -``````````` -.. doxygenfile:: paddle/cuda/include/hl_sparse.h - -hl_sparse.ph -```````````` -.. doxygenfile:: paddle/cuda/include/hl_sparse.ph diff --git a/doc/howto/source/cuda/nn.rst b/doc/howto/source/cuda/nn.rst deleted file mode 100644 index 5577d01e72..0000000000 --- a/doc/howto/source/cuda/nn.rst +++ /dev/null @@ -1,39 +0,0 @@ -Neural Network -============== - -Base ----- - -.. doxygenfile:: paddle/cuda/include/hl_gpu.h -.. doxygenfile:: paddle/cuda/include/hl_functions.h -.. doxygenfile:: paddle/cuda/include/hl_avx_functions.h -.. doxygenfile:: paddle/cuda/include/hl_gpu_functions.cuh -.. doxygenfile:: paddle/cuda/include/hl_activation_functions.h - - -CNN Related APIs ----------------- -.. doxygenfile:: paddle/cuda/include/hl_cnn.h -.. doxygenfile:: paddle/cuda/include/hl_cuda_cudnn.h -.. doxygenfile:: paddle/cuda/include/hl_cuda_cudnn.ph - -RNN Related APIs ----------------- - -.. doxygenfile:: paddle/cuda/include/hl_recurrent_apply.cuh -.. doxygenfile:: paddle/cuda/include/hl_sequence.h - -LSTM Model -`````````` - -.. doxygenfile:: paddle/cuda/include/hl_lstm.h -.. dpxygenfile:: paddle/cuda/include/hl_cpu_lstm.cuh -.. doxygenfile:: paddle/cuda/include/hl_gpu_lstm.cuh -.. doxygenfile:: paddle/cuda/include/hl_lstm_ops.cuh - -GRU Model -````````` - -.. doxygenfile:: paddle/cuda/include/hl_gru_ops.cuh -.. doxygenfile:: paddle/cuda/include/hl_cpu_gru.cuh -.. doxygenfile:: paddle/cuda/include/hl_gpu_gru.cuh diff --git a/doc/howto/source/cuda/utils.rst b/doc/howto/source/cuda/utils.rst deleted file mode 100644 index 850e8bd1c6..0000000000 --- a/doc/howto/source/cuda/utils.rst +++ /dev/null @@ -1,37 +0,0 @@ -Utils -===== - -Dynamic Link Libs ------------------ -.. doxygenfile:: paddle/cuda/include/hl_dso_loader.h - -GPU Resources -------------- - -hl_cuda.ph -`````````` -.. doxygenfile:: paddle/cuda/include/hl_cuda.ph - -hl_cuda.h -````````` -.. doxygenfile:: paddle/cuda/include/hl_cuda.h - -HPPL Base ---------- -.. doxygenfile:: paddle/cuda/include/hl_base.h - -CUBLAS Wrapper --------------- -.. doxygenfile:: paddle/cuda/include/hl_cuda_cublas.h - -Timer ------ -.. doxygenfile:: paddle/cuda/include/hl_time.h - -Thread Resource ---------------- -.. doxygenfile:: paddle/cuda/include/hl_thread.ph - -Device Function ---------------- -.. doxygenfile:: paddle/cuda/include/hl_device_functions.cuh diff --git a/doc/howto/source/gserver/activations.rst b/doc/howto/source/gserver/activations.rst deleted file mode 100644 index 55b9d3be38..0000000000 --- a/doc/howto/source/gserver/activations.rst +++ /dev/null @@ -1,5 +0,0 @@ -Activations -=========== - -.. doxygenclass:: paddle::ActivationFunction - :members: diff --git a/doc/howto/source/gserver/dataproviders.rst b/doc/howto/source/gserver/dataproviders.rst deleted file mode 100644 index c30d9d6a36..0000000000 --- a/doc/howto/source/gserver/dataproviders.rst +++ /dev/null @@ -1,87 +0,0 @@ -============== -Data Providers -============== - -DataProviders -============= - -Base ----- -.. doxygenclass:: paddle::DataProvider - :members: - -DataProviderGroup ------------------ -.. doxygenclass:: paddle::DataProviderGroup - :members: - -MultiDataProvider ------------------ -.. doxygenclass:: paddle::MultiDataProvider - :members: - -PyDataProvider -============== - -IFieldScanner -------------- -.. doxygenclass:: paddle::IFieldScanner - :members: - -DenseScanner -------------- -.. doxygenclass:: paddle::DenseScanner - :members: - -IndexScanner -------------- -.. doxygenclass:: paddle::IndexScanner - :members: - -SparseNonValueScanner ---------------------- -.. doxygenclass:: paddle::SparseNonValueScanner - :members: - -SparseValueScanner ------------------- -.. doxygenclass:: paddle::SparseValueScanner - :members: - -SequenceScanner ---------------- -.. doxygenclass:: paddle::SparseValueScanner - :members: - -IPyDataProviderCache --------------------- -.. doxygenclass:: paddle::IPyDataProviderCache - :members: - -NoCacheStrategy ---------------- -.. doxygenclass:: paddle::NoCacheStrategy - :members: - -CacheOnePassInMemory --------------------- -.. doxygenclass:: paddle::CacheOnePassInMemory - :members: - -IPyDataProvider ---------------- -.. doxygenclass:: paddle::PyDataProvider2 - :members: - -ProtoDataProvider -================= - -ProtoDataProvider ----------------- -.. doxygenclass:: paddle::ProtoDataProvider - :members: - -ProtoSequenceDataProvider -------------------------- -.. doxygenclass:: paddle::ProtoSequenceDataProvider - :members: diff --git a/doc/howto/source/gserver/evaluators.rst b/doc/howto/source/gserver/evaluators.rst deleted file mode 100644 index f5361f76cd..0000000000 --- a/doc/howto/source/gserver/evaluators.rst +++ /dev/null @@ -1,103 +0,0 @@ -========== -Evaluators -========== - -Base -==== - -.. doxygenclass:: paddle::Evaluator - :members: - -Sum -=== - -SumEvaluator ------------- -.. doxygenclass:: paddle::SumEvaluator - :members: - -ColumnSumEvaluator ------------------- -.. doxygenclass:: paddle::ColumnSumEvaluator - :members: - -Classification -============== - -ClassificationErrorEvaluator ---------------------------- -.. doxygenclass:: paddle::ClassificationErrorEvaluator - :members: - -SequenceClassificationErrorEvaluator ------------------------------------- -.. doxygenclass:: paddle::SequenceClassificationErrorEvaluator - :members: - -AucEvaluator -------------- -.. doxygenclass:: paddle::AucEvaluator - :members: - -PrecisionRecallEvaluator ------------------------- -.. doxygenclass:: paddle::PrecisionRecallEvaluator - :members: - -ChunkEvaluator --------------- -.. doxygenclass:: paddle::ChunkEvaluator - :members: - -CTCEvaluator ------------- -.. doxygenclass:: paddle::CTCErrorEvaluator - :members: - - -Rank -==== - -PnpairEvaluator -------------- -.. doxygenclass:: paddle::PnpairEvaluator - :members: - -AucEvaluator -------------- -.. doxygenclass:: paddle::RankAucEvaluator - :members: - - -Printer -======= - -ValuePrinter -------------- -.. doxygenclass:: paddle::ValuePrinter - :members: - -GradientPrinter ---------------- -.. doxygenclass:: paddle::GradientPrinter - :members: - -MaxIdPrinter ------------- -.. doxygenclass:: paddle::MaxIdPrinter - :members: - -MaxFramePrinter ---------------- -.. doxygenclass:: paddle::MaxFramePrinter - :members: - -SequenceTextPrinter ------------------- -.. doxygenclass:: paddle::SequenceTextPrinter - :members: - -ClassificationErrorPrinter --------------------------- -.. doxygenclass:: paddle::ClassificationErrorPrinter - :members: diff --git a/doc/howto/source/gserver/gradientmachines.rst b/doc/howto/source/gserver/gradientmachines.rst deleted file mode 100644 index 04c8e91d03..0000000000 --- a/doc/howto/source/gserver/gradientmachines.rst +++ /dev/null @@ -1,27 +0,0 @@ -Gradient Machines -================= - -GradientMachine ---------------- -.. doxygenclass:: paddle::GradientMachine - :members: - -GradientMachineMode -------------------- -.. doxygenclass:: paddle::IGradientMachineMode - :members: - -MultiGradientMachine --------------------- -.. doxygenclass:: paddle::MultiGradientMachine - :members: - -TrainerThread -````````````` -.. doxygenclass:: paddle::TrainerThread - :members: - -RecurrentGradientMachine ------------------------- -.. doxygenclass:: paddle::RecurrentGradientMachine - :members: diff --git a/doc/howto/source/gserver/index.rst b/doc/howto/source/gserver/index.rst deleted file mode 100644 index 223b00b9a9..0000000000 --- a/doc/howto/source/gserver/index.rst +++ /dev/null @@ -1,12 +0,0 @@ -GServer -======= - -.. toctree:: - :maxdepth: 2 - - activations.rst - dataproviders.rst - evaluators.rst - gradientmachines.rst - layers.rst - neworks.rst diff --git a/doc/howto/source/gserver/layers.rst b/doc/howto/source/gserver/layers.rst deleted file mode 100644 index 191b2bdff2..0000000000 --- a/doc/howto/source/gserver/layers.rst +++ /dev/null @@ -1,566 +0,0 @@ -====== -Layers -====== - -Base -==== - -Layer ------ -.. doxygenclass:: paddle::Layer - :members: - -Projection ----------- -.. doxygenclass:: paddle::Projection - :members: - -Operator --------- -.. doxygenclass:: paddle::Operator - :members: - -Data Layer -========== - -.. doxygenclass:: paddle::DataLayer - :members: - -Fully Connected Layers -====================== - -FullyConnectedLayer -------------------- -.. doxygenclass:: paddle::FullyConnectedLayer - :members: - -SelectiveFullyConnectedLayer ----------------------------- -.. doxygenclass:: paddle::SelectiveFullyConnectedLayer - :members: - -Conv Layers -=========== - -ConvBaseLayer -------------- -.. doxygenclass:: paddle::ConvBaseLayer - :members: - -ConvOperator ------------- -.. doxygenclass:: paddle::ConvOperator - :members: - -ConvShiftLayer --------------- -.. doxygenclass:: paddle::ConvShiftLayer - :members: - -CudnnConvLayer --------------- -.. doxygenclass:: paddle::CudnnConvLayer - :members: - -ExpandConvBaseLayer -------------------- -.. doxygenclass:: paddle::ExpandConvBaseLayer - :members: - -ExpandConvLayer ---------------- -.. doxygenclass:: paddle::ExpandConvLayer - :members: - -ContextProjection ------------------ -.. doxygenclass:: paddle::ContextProjection - :members: - -Pooling Layers -============== - -PoolLayer ---------- -.. doxygenclass:: paddle::PoolLayer - :members: - -PoolProjectionLayer -------------------- -.. doxygenclass:: paddle::PoolProjectionLayer - :members: - -CudnnPoolLayer --------------- -.. doxygenclass:: paddle::CudnnPoolLayer - :members: - -SpatialPyramidPoolLayer ------------------------ -.. doxygenclass:: paddle::SpatialPyramidPoolLayer - :members: - -MaxOutLayer ------------ -.. doxygenclass:: paddle::MaxOutLayer - :members: - -Norm Layers -=========== - -NormLayer ---------- -.. doxygenclass:: paddle::NormLayer - :members: - -CMRProjectionNormLayer ----------------------- -.. doxygenclass:: paddle::CMRProjectionNormLayer - :members: - -DataNormLayer -------------- -.. doxygenclass:: paddle::DataNormLayer - :members: - -ResponseNormLayer ------------------ -.. doxygenclass:: paddle::ResponseNormLayer - :members: - -BatchNormBaseLayer ------------------- -.. doxygenclass:: paddle::BatchNormBaseLayer - :members: - -BatchNormalizationLayer ------------------------ -.. doxygenclass:: paddle::BatchNormalizationLayer - :members: - -CudnnBatchNormLayer ------------------------ -.. doxygenclass:: paddle::CudnnBatchNormLayer - :members: - -SumToOneNormLayer ------------------ -.. doxygenclass:: paddle::SumToOneNormLayer - :members: - -Activation Layer -================ - -ParameterReluLayer ------------------- -.. doxygenclass:: paddle::ParameterReluLayer - :members: - -Recurrent Layers -================ - -RecurrentLayer --------------- -.. doxygenclass:: paddle::RecurrentLayer - :members: - -SequenceToBatch ---------------- -.. doxygenclass:: paddle::SequenceToBatch - :members: - -LSTM ----- -LstmLayer -````````` -.. doxygenclass:: paddle::LstmLayer - :members: - -LstmStepLayer -````````````` -.. doxygenclass:: paddle::LstmStepLayer - :members: - -LstmCompute -``````````` -.. doxygenclass:: paddle::LstmCompute - :members: - -MDLSTM ------- -MDLstmLayer -``````````` -.. doxygenclass:: paddle::MDLstmLayer - :members: - -CoordIterator -````````````` -.. doxygenclass:: paddle::CoordIterator - :members: - -GRU ---- -GatedRecurrentLayer -``````````````````` -.. doxygenclass:: paddle::GatedRecurrentLayer - :members: - -GruStepLayer -```````````` -.. doxygenclass:: paddle::GruStepLayer - :members: - -GruCompute -`````````` -.. doxygenclass:: paddle::GruCompute - :members: - -Recurrent Layer Group -===================== - -AgentLayer ----------- -.. doxygenclass:: paddle::AgentLayer - :members: - -SequenceAgentLayer ------------------- -.. doxygenclass:: paddle::SequenceAgentLayer - :members: - -GatherAgentLayer ----------------- -.. doxygenclass:: paddle::GatherAgentLayer - :members: - -SequenceGatherAgentLayer ------------------------- -.. doxygenclass:: paddle::SequenceGatherAgentLayer - :members: - -ScatterAgentLayer ------------------ -.. doxygenclass:: paddle::ScatterAgentLayer - :members: - -SequenceScatterAgentLayer -------------------------- -.. doxygenclass:: paddle::SequenceScatterAgentLayer - :members: - -GetOutputLayer --------------- -.. doxygenclass:: paddle::GetOutputLayer - :members: - -Mixed Layer -=========== -.. doxygenclass:: paddle::MixedLayer - :members: - -DotMulProjection ----------------- -.. doxygenclass:: paddle::DotMulProjection - :members: - -DotMulOperator --------------- -.. doxygenclass:: paddle::DotMulOperator - :members: - -FullMatrixProjection --------------------- -.. doxygenclass:: paddle::FullMatrixProjection - :members: - -IdentityProjection ------------------- -.. doxygenclass:: paddle::IdentityProjection - :members: - -IdentityOffsetProjection ------------------------- -.. doxygenclass:: paddle::IdentityOffsetProjection - :members: - -TableProjection ---------------- -.. doxygenclass:: paddle::TableProjection - :members: - -TransposedFullMatrixProjection ------------------------------- -.. doxygenclass:: paddle::TransposedFullMatrixProjection - :members: - -Aggregate Layers -================ - -Aggregate ---------- -AverageLayer -```````````` -.. doxygenclass:: paddle::AverageLayer - :members: - -MaxLayer -```````` -.. doxygenclass:: paddle::MaxLayer - :members: - -SequenceLastInstanceLayer -````````````````````````` -.. doxygenclass:: paddle::SequenceLastInstanceLayer - :members: - -Concat ------- -ConcatenateLayer -```````````````` -.. doxygenclass:: paddle::ConcatenateLayer - :members: - -ConcatenateLayer2 -````````````````` -.. doxygenclass:: paddle::ConcatenateLayer2 - :members: - -SequenceConcatLayer -``````````````````` -.. doxygenclass:: paddle::SequenceConcatLayer - :members: - -Subset ------- -SubSequenceLayer -```````````````` -.. doxygenclass:: paddle::SubSequenceLayer - :members: - -Reshaping Layers -================ - -BlockExpandLayer ----------------- -.. doxygenclass:: paddle::BlockExpandLayer - :members: - -ExpandLayer ------------ -.. doxygenclass:: paddle::ExpandLayer - :members: - -FeatureMapExpandLayer ---------------------- -.. doxygenclass:: paddle::FeatureMapExpandLayer - :members: - -ResizeLayer ------------ -.. doxygenclass:: paddle::ResizeLayer - :members: - -SequenceReshapeLayer --------------------- -.. doxygenclass:: paddle::SequenceReshapeLayer - :members: - -Math Layers -=========== - -AddtoLayer ----------- -.. doxygenclass:: paddle::AddtoLayer - :members: - -ConvexCombinationLayer ----------------------- -.. doxygenclass:: paddle::ConvexCombinationLayer - :members: - -InterpolationLayer ------------------- -.. doxygenclass:: paddle::InterpolationLayer - :members: - -MultiplexLayer --------------- -.. doxygenclass:: paddle::MultiplexLayer - :members: - -OuterProdLayer --------------- -.. doxygenclass:: paddle::OuterProdLayer - :members: - -PowerLayer ----------- -.. doxygenclass:: paddle::PowerLayer - :members: - -ScalingLayer ------------- -.. doxygenclass:: paddle::ScalingLayer - :members: - -SlopeInterceptLayer -------------------- -.. doxygenclass:: paddle::SlopeInterceptLayer - :members: - -TensorLayer ------------- -.. doxygenclass:: paddle::TensorLayer - :members: - -TransLayer ----------- -.. doxygenclass:: paddle::TransLayer - :members: - -Sampling Layers -=============== - -BilinearInterpLayer -------------------- -.. doxygenclass:: paddle::BilinearInterpLayer - :members: - -MultinomialSampler ------------------- -.. doxygenclass:: paddle::MultinomialSampler - :members: - -MaxIdLayer ----------- -.. doxygenclass:: paddle::MaxIdLayer - :members: - -SamplingIdLayer ---------------- -.. doxygenclass:: paddle::SamplingIdLayer - :members: - -Cost Layers -=========== - -CostLayer ------------ -.. doxygenclass:: paddle::CostLayer - :members: - -HuberTwoClass -````````````` -.. doxygenclass:: paddle::HuberTwoClass - :members: - -LambdaCost -``````````` -.. doxygenclass:: paddle::LambdaCost - :members: - -MultiBinaryLabelCrossEntropy -```````````````````````````` -.. doxygenclass:: paddle::MultiBinaryLabelCrossEntropy - :members: - -MultiClassCrossEntropy -``````````````````````` -.. doxygenclass:: paddle::MultiClassCrossEntropy - :members: - -MultiClassCrossEntropyWithSelfNorm -`````````````````````````````````` -.. doxygenclass:: paddle::MultiClassCrossEntropyWithSelfNorm - :members: - -RankingCost -``````````` -.. doxygenclass:: paddle::RankingCost - :members: - -SoftBinaryClassCrossEntropy -``````````````````````````` -.. doxygenclass:: paddle::SoftBinaryClassCrossEntropy - :members: - -SumOfSquaresCostLayer -````````````````````` -.. doxygenclass:: paddle::SumOfSquaresCostLayer - :members: - -SumCostLayer -````````````````````` -.. doxygenclass:: paddle::SumCostLayer - :members: - -CosSimLayer ------------ -.. doxygenclass:: paddle::CosSimLayer - :members: - -CosSimVecMatLayer ------------------ -.. doxygenclass:: paddle::CosSimVecMatLayer - :members: - -CRFDecodingLayer ----------------- -.. doxygenclass:: paddle::CRFDecodingLayer - :members: - -CRFLayer --------- -.. doxygenclass:: paddle::CRFLayer - :members: - -CTCLayer --------- -.. doxygenclass:: paddle::CTCLayer - :members: - -HierarchicalSigmoidLayer ------------------------- -.. doxygenclass:: paddle::HierarchicalSigmoidLayer - :members: - -LinearChainCRF --------------- -.. doxygenclass:: paddle::LinearChainCRF - :members: - -LinearChainCTC --------------- -.. doxygenclass:: paddle::LinearChainCTC - :members: - -NCELayer --------- -.. doxygenclass:: paddle::NCELayer - :members: - -Validation Layers ------------------ - -ValidationLayer -``````````````` -.. doxygenclass:: paddle::ValidationLayer - :members: - -AucValidation -````````````` -.. doxygenclass:: paddle::AucValidation - :members: - -PnpairValidation -```````````````` -.. doxygenclass:: paddle::PnpairValidation - :members: - -Check Layers -============ - -EosIdCheckLayer ---------------- -.. doxygenclass:: paddle::EosIdCheckLayer - :members: diff --git a/doc/howto/source/gserver/neworks.rst b/doc/howto/source/gserver/neworks.rst deleted file mode 100644 index 73fb60d549..0000000000 --- a/doc/howto/source/gserver/neworks.rst +++ /dev/null @@ -1,12 +0,0 @@ -Networks -======== - -NeuralNetwork -------------- -.. doxygenclass:: paddle::NeuralNetwork - :members: - -ParallelNeuralNetwork ---------------------- -.. doxygenclass:: paddle::ParallelNeuralNetwork - :members: diff --git a/doc/howto/source/index.rst b/doc/howto/source/index.rst deleted file mode 100644 index 36323c888e..0000000000 --- a/doc/howto/source/index.rst +++ /dev/null @@ -1,14 +0,0 @@ -Source Code Documents -===================== - -.. toctree:: - :maxdepth: 1 - - gserver/index.rst - trainer.rst - parameter/index.rst - pserver/index.rst - api.rst - cuda/index.rst - math/index.rst - utils/index.rst diff --git a/doc/howto/source/math/functions.rst b/doc/howto/source/math/functions.rst deleted file mode 100644 index aef12e0f00..0000000000 --- a/doc/howto/source/math/functions.rst +++ /dev/null @@ -1,10 +0,0 @@ -Functions -========= - -MathFunctions -------------- -.. doxygenfile:: paddle/math/MathFunctions.h - -SIMDFunctions -------------- -.. doxygenfile:: paddle/math/SIMDFunctions.h diff --git a/doc/howto/source/math/index.rst b/doc/howto/source/math/index.rst deleted file mode 100644 index 2ec16f2b44..0000000000 --- a/doc/howto/source/math/index.rst +++ /dev/null @@ -1,10 +0,0 @@ -Math -==== - -.. toctree:: - :maxdepth: 2 - - vector.rst - matrix.rst - functions.rst - utils.rst diff --git a/doc/howto/source/math/matrix.rst b/doc/howto/source/math/matrix.rst deleted file mode 100644 index 9bb20f618d..0000000000 --- a/doc/howto/source/math/matrix.rst +++ /dev/null @@ -1,76 +0,0 @@ -Matrix -====== - -Base ----- - -BaseMatrix Template -``````````````````` -.. doxygenclass:: paddle::BaseMatrixT - :members: - -Matrix -`````` -.. doxygenclass:: paddle::Matrix - :members: - -MatrixOffset -```````````` -.. doxygenclass:: paddle::MatrixOffset - :members: - -CpuMatrix ---------- - -CpuMatrix -````````` -.. doxygenclass:: paddle::CpuMatrix - :members: - -SharedCpuMatrix -``````````````` -.. doxygenclass:: paddle::SharedCpuMatrix - :members: - -GpuMatrix ---------- -.. doxygenclass:: paddle::GpuMatrix - :members: - -CpuSparseMatrix ---------------- - -CpuSparseMatrix -``````````````` -.. doxygenclass:: paddle::CpuSparseMatrix - :members: - -SparseRowCpuMatrix -`````````````````` -.. doxygenclass:: paddle::SparseRowCpuMatrix - :members: - -SparseAutoGrowRowCpuMatrix -`````````````````````````` -.. doxygenclass:: paddle::SparseAutoGrowRowCpuMatrix - :members: - -SparsePrefetchRowCpuMatrix -`````````````````````````` -.. doxygenclass:: paddle::SparsePrefetchRowCpuMatrix - :members: - -SparseRowIdsCpuMatrix -````````````````````` -.. doxygenclass:: paddle::SparseRowIdsCpuMatrix - :members: - -CacheRowCpuMatrix -````````````````` -.. doxygenclass:: paddle::CacheRowCpuMatrix - :members: - -GpuSparseMatrix ---------------- -.. doxygenclass:: paddle::GpuSparseMatrix - :members: diff --git a/doc/howto/source/math/utils.rst b/doc/howto/source/math/utils.rst deleted file mode 100644 index 55d9961a39..0000000000 --- a/doc/howto/source/math/utils.rst +++ /dev/null @@ -1,18 +0,0 @@ -Memory Manager -============== - -Memory Handle -------------- -.. doxygenfile:: paddle/math/MemoryHandle.h - -Allocator ---------- -.. doxygenfile:: paddle/math/Allocator.h - -PoolAllocator -````````````` -.. doxygenfile:: paddle/math/PoolAllocator.h - -Storage -------- -.. doxygenfile:: paddle/math/Storage.h diff --git a/doc/howto/source/math/vector.rst b/doc/howto/source/math/vector.rst deleted file mode 100644 index 07f7062aba..0000000000 --- a/doc/howto/source/math/vector.rst +++ /dev/null @@ -1,37 +0,0 @@ -Vector -====== - -BaseVector -`````````` -.. doxygenclass:: paddle::BaseVector - :members: - -Vector Template -``````````````` -.. doxygenclass:: paddle::VectorT - :members: - -CpuVector Template -`````````````````` -.. doxygenclass:: paddle::CpuVectorT - :members: - -GpuVector Template -`````````````````` -.. doxygenclass:: paddle::GpuVectorT - :members: - -ParallelCpuVector Template -`````````````````````````` -.. doxygenclass:: paddle::ParallelCpuVectorT - :members: - -ParallelGpuVector Template -`````````````````````````` -.. doxygenclass:: paddle::ParallelGpuVectorT - :members: - -CpuGpuVector Template -````````````````````` -.. doxygenclass:: paddle::CpuGpuVectorT - :members: diff --git a/doc/howto/source/parameter/index.rst b/doc/howto/source/parameter/index.rst deleted file mode 100644 index 3bf6948dc3..0000000000 --- a/doc/howto/source/parameter/index.rst +++ /dev/null @@ -1,9 +0,0 @@ -Parameter -========= - -.. toctree:: - :maxdepth: 2 - - parameter.rst - optimizer.rst - updater.rst diff --git a/doc/howto/source/parameter/optimizer.rst b/doc/howto/source/parameter/optimizer.rst deleted file mode 100644 index b5b8b850b3..0000000000 --- a/doc/howto/source/parameter/optimizer.rst +++ /dev/null @@ -1,22 +0,0 @@ -Optimizer -========= - -ParameterOptimizer ------------------- -.. doxygenfile:: paddle/parameter/ParameterOptimizer.h - -Regularizer ------------ -.. doxygenfile:: paddle/parameter/Regularizer.h - -FirstOrderOptimizer -------------------- -.. doxygenfile:: paddle/parameter/FirstOrderOptimizer.h - -AverageOptimizer ----------------- -.. doxygenfile:: paddle/parameter/AverageOptimizer.h - -OptimizerWithRegularizer ------------------------- -.. doxygenfile:: paddle/parameter/OptimizerWithRegularizer.h diff --git a/doc/howto/source/parameter/parameter.rst b/doc/howto/source/parameter/parameter.rst deleted file mode 100644 index 2daa62d4e6..0000000000 --- a/doc/howto/source/parameter/parameter.rst +++ /dev/null @@ -1,12 +0,0 @@ -Parameter -========= - -Parameter ---------- -.. doxygenfile:: paddle/parameter/Argument.h -.. doxygenfile:: paddle/parameter/Parameter.h -.. doxygenfile:: paddle/parameter/ParallelParameter.h - -Weight ------- -.. doxygenfile:: paddle/parameter/Weight.h diff --git a/doc/howto/source/parameter/updater.rst b/doc/howto/source/parameter/updater.rst deleted file mode 100644 index dfa22e8e7d..0000000000 --- a/doc/howto/source/parameter/updater.rst +++ /dev/null @@ -1,14 +0,0 @@ -Updater -======= - -Base ----- -.. doxygenfile:: paddle/parameter/ParameterUpdaterBase.h - -Hook ----- -.. doxygenfile:: paddle/parameter/ParameterUpdaterHook.h - -Functions ---------- -.. doxygenfile:: paddle/parameter/ParameterUpdateFunctions.h diff --git a/doc/howto/source/pserver/client.rst b/doc/howto/source/pserver/client.rst deleted file mode 100644 index e5bba0706a..0000000000 --- a/doc/howto/source/pserver/client.rst +++ /dev/null @@ -1,12 +0,0 @@ -Client -====== - -BaseClient ----------- -.. doxygenclass:: paddle::BaseClient - :members: - -ParameterClient2 ----------------- -.. doxygenclass:: paddle::ParameterClient2 - :members: diff --git a/doc/howto/source/pserver/index.rst b/doc/howto/source/pserver/index.rst deleted file mode 100644 index 0031e9476b..0000000000 --- a/doc/howto/source/pserver/index.rst +++ /dev/null @@ -1,10 +0,0 @@ -PServer -======= - -.. toctree:: - :maxdepth: 2 - - client.rst - network.rst - server.rst - utils.rst diff --git a/doc/howto/source/pserver/network.rst b/doc/howto/source/pserver/network.rst deleted file mode 100644 index 7004c9d91f..0000000000 --- a/doc/howto/source/pserver/network.rst +++ /dev/null @@ -1,27 +0,0 @@ -Network -======= - -SocketServer ------------- -.. doxygenclass:: paddle::SocketServer - :members: - -SocketWorker ------------- -.. doxygenclass:: paddle::SocketWorker - :members: - -SocketClient ------------- -.. doxygenclass:: paddle::SocketClient - :members: - -SocketChannel -------------- -.. doxygenclass:: paddle::SocketChannel - :members: - -MessageReader -------------- -.. doxygenclass:: paddle::MsgReader - :members: diff --git a/doc/howto/source/pserver/server.rst b/doc/howto/source/pserver/server.rst deleted file mode 100644 index 35301acf8f..0000000000 --- a/doc/howto/source/pserver/server.rst +++ /dev/null @@ -1,12 +0,0 @@ -Server -====== - -ProtoServer ------------ -.. doxygenclass:: paddle::ProtoServer - :members: - -ParameterServer2 ----------------- -.. doxygenclass:: paddle::ParameterServer2 - :members: diff --git a/doc/howto/source/trainer.rst b/doc/howto/source/trainer.rst deleted file mode 100644 index 85f1feb4fc..0000000000 --- a/doc/howto/source/trainer.rst +++ /dev/null @@ -1,32 +0,0 @@ -Trainer -======= - -TrainerStats ------------- - -.. doxygenclass:: paddle::TrainerStats - :members: - -RemoteParameterUpdater ------------------------ - -.. doxygenclass:: paddle::RemoteParameterUpdater - :members: - -ConcurrentRemoteParameterUpdater --------------------------------- - -.. doxygenclass:: paddle::ConcurrentRemoteParameterUpdater - :members: - -SparseRemoteParameterUpdater ----------------------------- - -.. doxygenclass:: paddle::SparseRemoteParameterUpdater - :members: - -SparseRemoteParameterUpdaterComposite -------------------------------------- - -.. doxygenclass:: paddle::SparseRemoteParameterUpdaterComposite - :members: diff --git a/doc/howto/source/utils/customStackTrace.rst b/doc/howto/source/utils/customStackTrace.rst deleted file mode 100644 index cdc8930739..0000000000 --- a/doc/howto/source/utils/customStackTrace.rst +++ /dev/null @@ -1,4 +0,0 @@ -CustomStackTrace -================ -.. doxygenclass:: paddle::CustomStackTrace - :members: diff --git a/doc/howto/source/utils/enum.rst b/doc/howto/source/utils/enum.rst deleted file mode 100644 index e0da75afe1..0000000000 --- a/doc/howto/source/utils/enum.rst +++ /dev/null @@ -1,3 +0,0 @@ -Enumeration wrapper -=================== -.. doxygennamespace:: paddle::enumeration_wrapper diff --git a/doc/howto/source/utils/index.rst b/doc/howto/source/utils/index.rst deleted file mode 100644 index 7ddc47d172..0000000000 --- a/doc/howto/source/utils/index.rst +++ /dev/null @@ -1,11 +0,0 @@ -Utils -===== - -.. toctree:: - :maxdepth: 2 - - lock.rst - queue.rst - thread.rst - customStackTrace.rst - enum.rst diff --git a/doc/howto/source/utils/lock.rst b/doc/howto/source/utils/lock.rst deleted file mode 100644 index f011acb943..0000000000 --- a/doc/howto/source/utils/lock.rst +++ /dev/null @@ -1,32 +0,0 @@ -Lock -==== - -RWLock ------- -.. doxygenclass:: paddle::RWLock - :members: - -ReadLockGuard -------------- -.. doxygenclass:: paddle::ReadLockGuard - :members: - -SpinLock --------- -.. doxygenclass:: paddle::SpinLock - :members: - -Semaphore ---------- -.. doxygenclass:: paddle::Semaphore - :members: - -ThreadBarrier -------------- -.. doxygenclass:: paddle::ThreadBarrier - :members: - -LockedCondition ---------------- -.. doxygenclass:: paddle::LockedCondition - :members: diff --git a/doc/howto/source/utils/queue.rst b/doc/howto/source/utils/queue.rst deleted file mode 100644 index 98192648e2..0000000000 --- a/doc/howto/source/utils/queue.rst +++ /dev/null @@ -1,12 +0,0 @@ -Queue -===== - -Queue ------ -.. doxygenclass:: paddle::Queue - :members: - -BlockingQueue -------------- -.. doxygenclass:: paddle::BlockingQueue - :members: diff --git a/doc/howto/source/utils/thread.rst b/doc/howto/source/utils/thread.rst deleted file mode 100644 index 23d379a989..0000000000 --- a/doc/howto/source/utils/thread.rst +++ /dev/null @@ -1,27 +0,0 @@ -Thread -====== - -Thread ------- -.. doxygenclass:: paddle::Thread - :members: - -ThreadWorker ------------- -.. doxygenclass:: paddle::ThreadWorker - :members: - -SyncThreadPool --------------- -.. doxygenclass:: paddle::SyncThreadPool - :members: - -MultiThreadWorker ------------------ -.. doxygenclass:: paddle::MultiThreadWorker - :members: - -AsyncThreadPool ---------------- -.. doxygenclass:: paddle::AsyncThreadPool - :members: diff --git a/paddle/scripts/docker/Dockerfile b/paddle/scripts/docker/Dockerfile index edb84712d8..6f0554caa5 100644 --- a/paddle/scripts/docker/Dockerfile +++ b/paddle/scripts/docker/Dockerfile @@ -11,7 +11,7 @@ RUN apt-get update \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y RUN pip install -U BeautifulSoup docopt PyYAML pillow \ - sphinx sphinx_rtd_theme breathe recommonmark + sphinx sphinx_rtd_theme recommonmark # cmake tends to hide and blur the dependencies between code modules, as # noted here https://github.com/PaddlePaddle/Paddle/issues/763. We are diff --git a/paddle/scripts/docker/Dockerfile.gpu b/paddle/scripts/docker/Dockerfile.gpu index 5d175e15a7..c707d3b553 100644 --- a/paddle/scripts/docker/Dockerfile.gpu +++ b/paddle/scripts/docker/Dockerfile.gpu @@ -11,7 +11,7 @@ RUN apt-get update \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y RUN pip install -U BeautifulSoup docopt PyYAML pillow \ - sphinx sphinx_rtd_theme breathe recommonmark + sphinx sphinx_rtd_theme recommonmark # cmake tends to hide and blur the dependencies between code modules, as # noted here https://github.com/PaddlePaddle/Paddle/issues/763. We are diff --git a/paddle/scripts/tools/build_docs/Dockerfile b/paddle/scripts/tools/build_docs/Dockerfile index 506b13210b..78dc756bd1 100644 --- a/paddle/scripts/tools/build_docs/Dockerfile +++ b/paddle/scripts/tools/build_docs/Dockerfile @@ -3,5 +3,5 @@ COPY build.sh / RUN pip install sphinx &&\ pip install sphinx_rtd_theme &&\ apt install -y doxygen graphviz &&\ - pip install breathe recommonmark numpy protobuf==2.6.1 + pip install recommonmark numpy protobuf==2.6.1 CMD /build.sh From b9579a5f01ee8fcf28c18081f65fe8d2b95cb03a Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Fri, 9 Dec 2016 17:27:41 +0800 Subject: [PATCH 39/60] Fix dead link after rearrange english documentation. * Fix #803 --- doc/api/data_provider/pydataprovider2_en.rst | 4 +- doc/api/index_en.rst | 32 ++++++++-------- .../trainer_config_helpers/data_sources.rst | 2 + doc/api/trainer_config_helpers/layers.rst | 12 ++++++ doc/api/trainer_config_helpers/networks.rst | 2 + doc/conf.py.in | 1 + doc/howto/cmd_parameter/index_en.md | 3 ++ doc/tutorials/rec/ml_dataset_en.md | 5 +++ doc/tutorials/rec/ml_regression_en.rst | 37 +++++++------------ 9 files changed, 58 insertions(+), 40 deletions(-) diff --git a/doc/api/data_provider/pydataprovider2_en.rst b/doc/api/data_provider/pydataprovider2_en.rst index b42cbca576..083436e271 100644 --- a/doc/api/data_provider/pydataprovider2_en.rst +++ b/doc/api/data_provider/pydataprovider2_en.rst @@ -1,5 +1,7 @@ +.. _api_pydataprovider: + PyDataProvider2 -================= +=============== We highly recommand users to use PyDataProvider2 to provide training or testing data to PaddlePaddle. The user only needs to focus on how to read a single diff --git a/doc/api/index_en.rst b/doc/api/index_en.rst index 9930f93e10..6fdee9f928 100644 --- a/doc/api/index_en.rst +++ b/doc/api/index_en.rst @@ -1,35 +1,37 @@ API -==== +=== DataProvider API ---------------- .. toctree:: - :maxdepth: 1 + :maxdepth: 1 - data_provider/index_en.rst - data_provider/pydataprovider2_en.rst + data_provider/index_en.rst + data_provider/pydataprovider2_en.rst + +.. _api_trainer_config: Model Config API ---------------- .. toctree:: - :maxdepth: 1 + :maxdepth: 1 - trainer_config_helpers/optimizers.rst - trainer_config_helpers/data_sources.rst - trainer_config_helpers/layers.rst - trainer_config_helpers/activations.rst - trainer_config_helpers/poolings.rst - trainer_config_helpers/networks.rst - trainer_config_helpers/evaluators.rst - trainer_config_helpers/attrs.rst + trainer_config_helpers/optimizers.rst + trainer_config_helpers/data_sources.rst + trainer_config_helpers/layers.rst + trainer_config_helpers/activations.rst + trainer_config_helpers/poolings.rst + trainer_config_helpers/networks.rst + trainer_config_helpers/evaluators.rst + trainer_config_helpers/attrs.rst Applications API ---------------- .. toctree:: - :maxdepth: 1 + :maxdepth: 1 - predict/swig_py_paddle_en.rst + predict/swig_py_paddle_en.rst diff --git a/doc/api/trainer_config_helpers/data_sources.rst b/doc/api/trainer_config_helpers/data_sources.rst index 44ea59df43..b9dd4dda01 100644 --- a/doc/api/trainer_config_helpers/data_sources.rst +++ b/doc/api/trainer_config_helpers/data_sources.rst @@ -1,3 +1,5 @@ +.. _api_trainer_config_helpers_data_sources: + DataSources =========== diff --git a/doc/api/trainer_config_helpers/layers.rst b/doc/api/trainer_config_helpers/layers.rst index b487b739a7..12a75080d0 100644 --- a/doc/api/trainer_config_helpers/layers.rst +++ b/doc/api/trainer_config_helpers/layers.rst @@ -20,6 +20,8 @@ LayerOutput Data layer =========== +.. _api_trainer_config_helpers_layers_data_layer: + data_layer ---------- .. automodule:: paddle.trainer_config_helpers.layers @@ -29,6 +31,8 @@ data_layer Fully Connected Layers ====================== +.. _api_trainer_config_helpers_layers_fc_layer: + fc_layer -------- .. automodule:: paddle.trainer_config_helpers.layers @@ -68,6 +72,8 @@ img_conv_layer :members: img_conv_layer :noindex: +.. _api_trainer_config_helpers_layers_context_projection: + context_projection ------------------ .. automodule:: paddle.trainer_config_helpers.layers @@ -185,6 +191,8 @@ mixed_layer :members: mixed_layer :noindex: +.. _api_trainer_config_helpers_layers_embedding_layer: + embedding_layer --------------- .. automodule:: paddle.trainer_config_helpers.layers @@ -237,6 +245,8 @@ trans_full_matrix_projection Aggregate Layers ================ +.. _api_trainer_config_helpers_layers_pooling_layer: + pooling_layer ------------- .. automodule:: paddle.trainer_config_helpers.layers @@ -333,6 +343,8 @@ tensor_layer :members: tensor_layer :noindex: +.. _api_trainer_config_helpers_layers_cos_sim: + cos_sim ------- .. automodule:: paddle.trainer_config_helpers.layers diff --git a/doc/api/trainer_config_helpers/networks.rst b/doc/api/trainer_config_helpers/networks.rst index 29c52c5ce3..e13c368051 100644 --- a/doc/api/trainer_config_helpers/networks.rst +++ b/doc/api/trainer_config_helpers/networks.rst @@ -13,6 +13,8 @@ sequence_conv_pool :members: sequence_conv_pool :noindex: +.. _api_trainer_config_helpers_network_text_conv_pool: + text_conv_pool -------------- .. automodule:: paddle.trainer_config_helpers.networks diff --git a/doc/conf.py.in b/doc/conf.py.in index 5fb307e3a9..01d156e887 100644 --- a/doc/conf.py.in +++ b/doc/conf.py.in @@ -144,5 +144,6 @@ def setup(app): # no c++ API for now app.add_config_value('recommonmark_config', { 'url_resolver': lambda url: github_doc_root + url, + 'enable_eval_rst': True, }, True) app.add_transform(AutoStructify) diff --git a/doc/howto/cmd_parameter/index_en.md b/doc/howto/cmd_parameter/index_en.md index bd16affdd8..a6c236db61 100644 --- a/doc/howto/cmd_parameter/index_en.md +++ b/doc/howto/cmd_parameter/index_en.md @@ -1,3 +1,6 @@ +```eval_rst +.. _cmd_line_index: +``` # How to Set Command-line Parameters * [Use Case](use_case_en.md) diff --git a/doc/tutorials/rec/ml_dataset_en.md b/doc/tutorials/rec/ml_dataset_en.md index c93a4585e4..73879d6537 100644 --- a/doc/tutorials/rec/ml_dataset_en.md +++ b/doc/tutorials/rec/ml_dataset_en.md @@ -1,3 +1,8 @@ +```eval_rst +.. _demo_ml_dataset: + +``` + # MovieLens Dataset The [MovieLens Dataset](http://grouplens.org/datasets/movielens/) was collected by GroupLens Research. diff --git a/doc/tutorials/rec/ml_regression_en.rst b/doc/tutorials/rec/ml_regression_en.rst index 0c14e4f5bb..b9c2f8cb59 100644 --- a/doc/tutorials/rec/ml_regression_en.rst +++ b/doc/tutorials/rec/ml_regression_en.rst @@ -16,7 +16,7 @@ Data Preparation ```````````````` Download and extract dataset '''''''''''''''''''''''''''' -We use `movielens 1m dataset `_ here. +We use :ref:`demo_ml_dataset` here. To download and unzip the dataset, simply run the following commands. .. code-block:: bash @@ -239,26 +239,16 @@ Then we combine each features of movie into one movie feature by a get one user feature. Then we calculate the cosine similarity of these two features. -In these network, we use several api in `trainer_config_helpers -<../../ui/api/trainer_config_helpers/index.html>`_. There are - -* Data Layer, `data_layer - <../../ui/api/trainer_config_helpers/layers.html#id1>`_ -* Fully Connected Layer, `fc_layer - <../../ui/api/trainer_config_helpers/layers.html#fc-layer>`_ -* Embedding Layer, `embedding_layer - <../../ui/api/trainer_config_helpers/layers.html#embedding-layer>`_ -* Context Projection Layer, `context_projection - <../../ui/api/trainer_config_helpers/layers.html#context-projection>`_ -* Pooling Layer, `pooling_layer - <../../ui/api/trainer_config_helpers/layers.html#pooling-layer>`_ -* Cosine Similarity Layer, `cos_sim - <../../ui/api/trainer_config_helpers/layers.html#cos-sim>`_ -* Text Convolution Pooling Layer, `text_conv_pool - <../../ui/api/trainer_config_helpers/networks.html - #trainer_config_helpers.networks.text_conv_pool>`_ -* Declare Python Data Sources, `define_py_data_sources2 - <../../ui/api/trainer_config_helpers/data_sources.html>`_ +In these network, we use several api in :ref:`api_trainer_config` . There are + +* Data Layer, :ref:`api_trainer_config_helpers_layers_data_layer` +* Fully Connected Layer, :ref:`api_trainer_config_helpers_layers_fc_layer` +* Embedding Layer, :ref:`api_trainer_config_helpers_layers_embedding_layer` +* Context Projection Layer, :ref:`api_trainer_config_helpers_layers_context_projection` +* Pooling Layer, :ref:`api_trainer_config_helpers_layers_pooling_layer` +* Cosine Similarity Layer, :ref:`api_trainer_config_helpers_layers_cos_sim` +* Text Convolution Pooling Layer, :ref:`api_trainer_config_helpers_network_text_conv_pool` +* Declare Python Data Sources :ref:`api_trainer_config_helpers_data_sources`. Data Provider ''''''''''''' @@ -274,7 +264,7 @@ In this :code:`dataprovider.py`, we should set\: * use_seq\: Whether this :code:`dataprovider.py` in sequence mode or not. * process\: Return each sample of data to :code:`paddle`. -The data provider details document see `there <../../ui/data_provider/pydataprovider2.html>`_. +The data provider details document see :ref:`api_pydataprovider`. Train ````` @@ -290,8 +280,7 @@ The run.sh is shown as follow: It just start a paddle training process, write the log to `log.txt`, then print it on screen. -Each command line argument in :code:`run.sh`, please refer to the `command line -arguments <../../ui/index.html#command-line-argument>`_ page. The short description of these arguments is shown as follow. +Each command line argument in :code:`run.sh`, please refer to the :ref:`cmd_line_index` page. The short description of these arguments is shown as follow. * config\: Tell paddle which file is neural network configuration. * save_dir\: Tell paddle save model into './output' From b8595196292378c596e4663a843612573588a0e4 Mon Sep 17 00:00:00 2001 From: livc Date: Fri, 9 Dec 2016 17:39:04 +0800 Subject: [PATCH 40/60] change filename to index_cn.md --- .../semantic_role_labeling/index_cn.md | 201 ++++++++++++++++++ 1 file changed, 201 insertions(+) create mode 100644 doc/tutorials/semantic_role_labeling/index_cn.md diff --git a/doc/tutorials/semantic_role_labeling/index_cn.md b/doc/tutorials/semantic_role_labeling/index_cn.md new file mode 100644 index 0000000000..f3c855a9fd --- /dev/null +++ b/doc/tutorials/semantic_role_labeling/index_cn.md @@ -0,0 +1,201 @@ +# 语义角色标注教程 # + +语义角色标注（Semantic role labeling, SRL）是浅语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词参数结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: + + [ A0 他 ] [ AM-MOD 将 ][ AM-NEG 不会 ] [ V 接受] [ A1 任何东西 ] 从 [A2 那些他写的东西中 ]。 + +- V: 动词 +- A0: 接受者 +- A1: 接受的东西 +- A2: 从……接受 +- A3: 属性 +- AM-MOD: 情态动词 +- AM-NEG: 否定 + +给定动词“接受”，句子中的大部分将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 + +到目前为止，大多数成功的SRL系统是建立在某种形式的解析结果之上的，其中在语法结构上使用了预先定义的特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标记问题。 + +## 数据描述 +相关论文[2]采用 CoNLL-2005＆2012 共享任务中设置的数据进行训练和测试。根据数据许可证，演示采用 CoNLL-2005 的测试数据集，可以在网站上找到。 + +用户只需执行以下命令就可以下载并处理原始数据： + +```bash +cd data +./get_data.sh +``` +`data `目录会出现如下几个新的文件： +```bash +conll05st-release：the test data set of CoNll-2005 shared task +test.wsj.words：the Wall Street Journal data sentences +test.wsj.props: the propositional arguments +feature: the extracted features from data set +``` + +## 训练 +### DB-LSTM +请参阅情绪分析的演示以了解有关长期短期记忆单元的更多信息。 + +与在 Sentiment Analysis 演示中使用的 Bidirectional-LSTM 不同，DB-LSTM 采用另一种方法来堆叠LSTM层。首先，标准LSTM以正向处理该序列。该 LSTM 层的输入和输出作为下一个 LSTM 层的输入，并被反向处理。这两个标准 LSTM 层组成一对 LSTM。然后我们堆叠一对对的 LSTM 层后得到深度 LSTM 模型。 + +下图展示了时间扩展的2层 DB-LSTM 网络。 +
+![pic](./network_arch.png) +
+ +### 特征 +两个输入特性在这个管道中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示： +
+![pic](./feature.jpg) +
+ +在这个示例中，相应的标记句子是： + +[ A1 A record date ] has [ AM-NEG n't ] been [ V set ] . + +在演示中, 我们采用上面的特征模板, 包括： `argument`, `predicate`, `ctx-p (p=-1,0,1)`, `mark` 并使用 `B/I/O` 方案来标记每个参数。这些特征和标签存储在 `feature` 文件中, 用`\t`分割。 + +### 数据提供 + +`dataprovider.py` 是一个包装数据的 Python 文件。 函数 `hook()` 定义了网络的数据槽。六个特征和标签都是索引槽。 +``` +def hook(settings, word_dict, label_dict, **kwargs): + settings.word_dict = word_dict + settings.label_dict = label_dict + #all inputs are integral and sequential type + settings.slots = [ + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(predicate_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(word_dict)), + integer_value_sequence(2), + integer_value_sequence(len(label_dict))] +``` +相应的数据迭代器如下： +``` +@provider(init_hook=hook, should_shuffle=True, calc_batch_size=get_batch_size, + can_over_batch_size=False, cache=CacheType.CACHE_PASS_IN_MEM) +def process(settings, file_name): + with open(file_name, 'r') as fdata: + for line in fdata: + sentence, predicate, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2, mark, label = \ + line.strip().split('\t') + + words = sentence.split() + sen_len = len(words) + word_slot = [settings.word_dict.get(w, UNK_IDX) for w in words] + + predicate_slot = [settings.predicate_dict.get(predicate)] * sen_len + ctx_n2_slot = [settings.word_dict.get(ctx_n2, UNK_IDX)] * sen_len + ctx_n1_slot = [settings.word_dict.get(ctx_n1, UNK_IDX)] * sen_len + ctx_0_slot = [settings.word_dict.get(ctx_0, UNK_IDX)] * sen_len + ctx_p1_slot = [settings.word_dict.get(ctx_p1, UNK_IDX)] * sen_len + ctx_p2_slot = [settings.word_dict.get(ctx_p2, UNK_IDX)] * sen_len + + marks = mark.split() + mark_slot = [int(w) for w in marks] + + label_list = label.split() + label_slot = [settings.label_dict.get(w) for w in label_list] + yield word_slot, predicate_slot, ctx_n2_slot, ctx_n1_slot, \ + ctx_0_slot, ctx_p1_slot, ctx_p2_slot, mark_slot, label_slot +``` +函数 `process` 产出有8个特征和标签的9个表。 + +### 神经网络配置 + +`db_lstm.py` 是在训练过程中加载字典并定义数据提供程序模块和网络架构的神经网络配置文件。 + +九个 `data_layer` 从数据提供程序加载实例。八个特征分别转换为嵌入，并由`mixed_layer`混合。 深度双向LSTM层提取softmax层的特征。目标函数是标签的交叉熵。 + +### 训练 +训练的脚本是 `train.sh`，用户只需执行: +```bash + ./train.sh +``` +`train.sh` 中的内容： +``` +paddle train \ + --config=./db_lstm.py \ + --use_gpu=0 \ + --log_period=5000 \ + --trainer_count=1 \ + --show_parameter_stats_period=5000 \ + --save_dir=./output \ + --num_passes=10000 \ + --average_test_period=10000000 \ + --init_model_path=./data \ + --load_missing_parameter_strategy=rand \ + --test_all_data_in_one_period=1 \ +2>&1 | tee 'train.log' +``` + +- \--config=./db_lstm.py : 网络配置文件 +- \--use_gpu=false: 使用 CPU 训练（如果已安装 PaddlePaddle GPU版本并想使用 GPU 训练可以设置为true，目前 crf_layer 不支持 GPU） +- \--log_period=500: 每20批(batch)输出日志 +- \--trainer_count=1: 设置线程数（或 GPU 数） +- \--show_parameter_stats_period=5000: 每100批显示参数统计 +- \--save_dir=./output: 模型输出路径 +- \--num_passes=10000: 设置通过数，一次通过意味着PaddlePaddle训练数据集中的所有样本一次 +- \--average_test_period=10000000: 每个 average_test_period 批次对平均参数进行测试 +- \--init_model_path=./data: 参数初始化路径 +- \--load_missing_parameter_strategy=rand: 随机初始不存在的参数 +- \--test_all_data_in_one_period=1: 在一个周期内测试所有数据 + + +训练后，模型将保存在目录`output`中。 我们的训练曲线如下： +
+![pic](./curve.jpg) +
+ +### 测试 +测试脚本是 `test.sh`, 执行: +```bash + ./test.sh +``` +`tesh.sh` 的主要部分： +``` +paddle train \ + --config=./db_lstm.py \ + --model_list=$model_list \ + --job=test \ + --config_args=is_test=1 \ +``` + + - \--config=./db_lstm.py: 网络配置文件 + - \--model_list=$model_list.list: 模型列表文件 + - \--job=test: 指示测试任务 + - \--config_args=is_test=1: 指示测试任务的标记 + - \--test_all_data_in_one_period=1: 在一个周期内测试所有数据 + + +### 预测 +预测脚本是 `predict.sh`，用户只需执行： +```bash + ./predict.sh + +``` +在`predict.sh`中，用户应该提供网络配置文件，模型路径，标签文件，字典文件，特征文件。 +``` +python predict.py + -c $config_file \ + -w $best_model_path \ + -l $label_file \ + -p $predicate_dict_file \ + -d $dict_file \ + -i $input_file \ + -o $output_file +``` + +`predict.py` 是主要的可执行python脚本，其中包括函数：加载模型，加载数据，数据预测。网络模型将输出标签的概率分布。 在演示中，我们使用最大概率的标签作为结果。用户还可以根据概率分布矩阵实现集束搜索或维特比解码。 + +预测后，结果保存在 `predict.res` 中。 + +## 引用 +[1] Martha Palmer, Dan Gildea, and Paul Kingsbury. The Proposition Bank: An Annotated Corpus of Semantic Roles , Computational Linguistics, 31(1), 2005. + +[2] Zhou, Jie, and Wei Xu. "End-to-end learning of semantic role labeling using recurrent neural networks." Proceedings of the Annual Meeting of the Association for Computational Linguistics. 2015. From aa0e415f96cb2aa95a0bb64cd68b58bd1218f7fc Mon Sep 17 00:00:00 2001 From: livc Date: Fri, 9 Dec 2016 17:57:35 +0800 Subject: [PATCH 41/60] Modify the translation details --- .../semantic_role_labeling/index_cn.md | 26 +++++++++---------- 1 file changed, 13 insertions(+), 13 deletions(-) diff --git a/doc/tutorials/semantic_role_labeling/index_cn.md b/doc/tutorials/semantic_role_labeling/index_cn.md index f3c855a9fd..7a3eff90e3 100644 --- a/doc/tutorials/semantic_role_labeling/index_cn.md +++ b/doc/tutorials/semantic_role_labeling/index_cn.md @@ -1,8 +1,8 @@ # 语义角色标注教程 # -语义角色标注（Semantic role labeling, SRL）是浅语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词参数结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: +语义角色标注（Semantic role labeling, SRL）是浅层语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词参数结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: - [ A0 他 ] [ AM-MOD 将 ][ AM-NEG 不会 ] [ V 接受] [ A1 任何东西 ] 从 [A2 那些他写的东西中 ]。 + [ A0 He ] [ AM-MOD would ][ AM-NEG n’t ] [ V accept] [ A1 anything of value ] from [A2 those he was writing about ]. - V: 动词 - A0: 接受者 @@ -12,12 +12,12 @@ - AM-MOD: 情态动词 - AM-NEG: 否定 -给定动词“接受”，句子中的大部分将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 +给定动词“accept”，句子中的大部分将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 -到目前为止，大多数成功的SRL系统是建立在某种形式的解析结果之上的，其中在语法结构上使用了预先定义的特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标记问题。 +到目前为止，大多数成功的SRL系统是建立在某种形式的解析结果之上的，其中在语法结构上使用了预先定义的特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标注问题。 ## 数据描述 -相关论文[2]采用 CoNLL-2005＆2012 共享任务中设置的数据进行训练和测试。根据数据许可证，演示采用 CoNLL-2005 的测试数据集，可以在网站上找到。 +相关论文[2]采用 CoNLL-2005＆2012 共享任务中设置的数据进行训练和测试。由于数据许可的原因，演示采用 CoNLL-2005 的测试数据集，可以在网站上找到。 用户只需执行以下命令就可以下载并处理原始数据： @@ -35,7 +35,7 @@ feature: the extracted features from data set ## 训练 ### DB-LSTM -请参阅情绪分析的演示以了解有关长期短期记忆单元的更多信息。 +请参阅情感分析的演示以了解有关长期短期记忆单元的更多信息。 与在 Sentiment Analysis 演示中使用的 Bidirectional-LSTM 不同，DB-LSTM 采用另一种方法来堆叠LSTM层。首先，标准LSTM以正向处理该序列。该 LSTM 层的输入和输出作为下一个 LSTM 层的输入，并被反向处理。这两个标准 LSTM 层组成一对 LSTM。然后我们堆叠一对对的 LSTM 层后得到深度 LSTM 模型。 @@ -45,7 +45,7 @@ feature: the extracted features from data set ### 特征 -两个输入特性在这个管道中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示： +两个输入特性在这个流程中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示：
![pic](./feature.jpg)
@@ -104,13 +104,13 @@ def process(settings, file_name): yield word_slot, predicate_slot, ctx_n2_slot, ctx_n1_slot, \ ctx_0_slot, ctx_p1_slot, ctx_p2_slot, mark_slot, label_slot ``` -函数 `process` 产出有8个特征和标签的9个表。 +函数 `process` 返回8个特征list和1个标签list。 ### 神经网络配置 `db_lstm.py` 是在训练过程中加载字典并定义数据提供程序模块和网络架构的神经网络配置文件。 -九个 `data_layer` 从数据提供程序加载实例。八个特征分别转换为嵌入，并由`mixed_layer`混合。 深度双向LSTM层提取softmax层的特征。目标函数是标签的交叉熵。 +九个 `data_layer` 从数据提供程序加载实例。八个特征分别转换为向量，并由`mixed_layer`混合。 深度双向LSTM层提取softmax层的特征。目标函数是标签的交叉熵。 ### 训练 训练的脚本是 `train.sh`，用户只需执行: @@ -136,11 +136,11 @@ paddle train \ - \--config=./db_lstm.py : 网络配置文件 - \--use_gpu=false: 使用 CPU 训练（如果已安装 PaddlePaddle GPU版本并想使用 GPU 训练可以设置为true，目前 crf_layer 不支持 GPU） -- \--log_period=500: 每20批(batch)输出日志 +- \--log_period=500: 每20个batch输出日志 - \--trainer_count=1: 设置线程数（或 GPU 数） -- \--show_parameter_stats_period=5000: 每100批显示参数统计 +- \--show_parameter_stats_period=5000: 每100个batch显示参数统计 - \--save_dir=./output: 模型输出路径 -- \--num_passes=10000: 设置通过数，一次通过意味着PaddlePaddle训练数据集中的所有样本一次 +- \--num_passes=10000: 设置数据遍历次数，一个pass意味着PaddlePaddle训练数据集中的所有样本被遍历一次 - \--average_test_period=10000000: 每个 average_test_period 批次对平均参数进行测试 - \--init_model_path=./data: 参数初始化路径 - \--load_missing_parameter_strategy=rand: 随机初始不存在的参数 @@ -191,7 +191,7 @@ python predict.py -o $output_file ``` -`predict.py` 是主要的可执行python脚本，其中包括函数：加载模型，加载数据，数据预测。网络模型将输出标签的概率分布。 在演示中，我们使用最大概率的标签作为结果。用户还可以根据概率分布矩阵实现集束搜索或维特比解码。 +`predict.py` 是主要的可执行python脚本，其中包括函数：加载模型，加载数据，数据预测。网络模型将输出标签的概率分布。 在演示中，我们使用最大概率的标签作为结果。用户还可以根据概率分布矩阵实现柱搜索或维特比解码。 预测后，结果保存在 `predict.res` 中。 From 9aa88426e1d78debe4853a3ac909868f590e9b0c Mon Sep 17 00:00:00 2001 From: zhangjcqq <664122220@qq.com> Date: Fri, 9 Dec 2016 19:02:43 +0800 Subject: [PATCH 42/60] modify somewhere --- doc/tutorials/semantic_role_labeling/index_cn.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/doc/tutorials/semantic_role_labeling/index_cn.md b/doc/tutorials/semantic_role_labeling/index_cn.md index 7a3eff90e3..c7e0a78f50 100644 --- a/doc/tutorials/semantic_role_labeling/index_cn.md +++ b/doc/tutorials/semantic_role_labeling/index_cn.md @@ -1,6 +1,6 @@ # 语义角色标注教程 # -语义角色标注（Semantic role labeling, SRL）是浅层语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词参数结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: +语义角色标注（Semantic role labeling, SRL）是浅层语义解析的一种形式，其目的是在给定的输入句子中发现每个谓词的谓词论元结构。 SRL作为很多自然语言处理任务中的中间步骤是很有用的，如信息提取、文档自动分类和问答。 实例如下 [1]: [ A0 He ] [ AM-MOD would ][ AM-NEG n’t ] [ V accept] [ A1 anything of value ] from [A2 those he was writing about ]. @@ -12,9 +12,9 @@ - AM-MOD: 情态动词 - AM-NEG: 否定 -给定动词“accept”，句子中的大部分将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 +给定动词“accept”，句子中的组块将会扮演某些语义角色。这里，标签方案来自 Penn Proposition Bank。 -到目前为止，大多数成功的SRL系统是建立在某种形式的解析结果之上的，其中在语法结构上使用了预先定义的特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标注问题。 +到目前为止，大多数成功的SRL系统是建立在某种形式的句法分析结果之上的，使用了基于句法结构的预定义特征模板。 本教程将介绍使用深度双向长短期记忆（DB-LSTM）模型[2]的端到端系统来解决SRL任务，这在很大程度上优于先前的最先进的系统。 这个系统将SRL任务视为序列标注问题。 ## 数据描述 相关论文[2]采用 CoNLL-2005＆2012 共享任务中设置的数据进行训练和测试。由于数据许可的原因，演示采用 CoNLL-2005 的测试数据集，可以在网站上找到。 @@ -45,7 +45,7 @@ feature: the extracted features from data set ### 特征 -两个输入特性在这个流程中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示： +两个输入特征在这个流程中起着至关重要的作用：predicate（pred）和argument（arguments）。 还采用了两个其他特征：谓词上下文（ctx-p）和区域标记（mr）。 因为单个谓词不能精确地描述谓词信息，特别是当相同的词在句子中出现多于一次时。 使用谓词上下文，可以在很大程度上消除歧义。类似地，如果它位于谓词上下文区域中，则使用区域标记 mr = 1 来表示参数位置，反之则 mr = 0。这四个简单的特征是我们的SRL系统所需要的。上下文大小设置为1的一个样本的特征如下[2]所示：
![pic](./feature.jpg)
From 56b15bd530a448e44264ad7b7643ead70c2f56c6 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sat, 10 Dec 2016 09:25:18 -0800 Subject: [PATCH 43/60] Install libgtest-dev in Dockerfiles --- paddle/scripts/docker/Dockerfile | 4 +++- paddle/scripts/docker/Dockerfile.gpu | 4 +++- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/paddle/scripts/docker/Dockerfile b/paddle/scripts/docker/Dockerfile index edb84712d8..6243d1d188 100644 --- a/paddle/scripts/docker/Dockerfile +++ b/paddle/scripts/docker/Dockerfile @@ -4,12 +4,14 @@ MAINTAINER PaddlePaddle Authors ARG DEBIAN_FRONTEND=noninteractive RUN apt-get update \ && apt-get install -y cmake libprotobuf-dev protobuf-compiler git \ - libgoogle-glog-dev libgflags-dev libatlas-dev libatlas3-base g++ m4 python-pip \ + libgoogle-glog-dev libgflags-dev libgtest-dev \ + libatlas-dev libatlas3-base g++ m4 python-pip \ python-protobuf python-numpy python-dev swig openssh-server \ wget unzip python-matplotlib tar xz-utils bzip2 gzip coreutils \ sed grep graphviz libjpeg-dev zlib1g-dev doxygen \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y +RUN cd /usr/src/gtest && cmake . && make && sudo cp *.a /usr/lib RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark diff --git a/paddle/scripts/docker/Dockerfile.gpu b/paddle/scripts/docker/Dockerfile.gpu index 5d175e15a7..f9821f2760 100644 --- a/paddle/scripts/docker/Dockerfile.gpu +++ b/paddle/scripts/docker/Dockerfile.gpu @@ -4,12 +4,14 @@ MAINTAINER PaddlePaddle Authors ARG DEBIAN_FRONTEND=noninteractive RUN apt-get update \ && apt-get install -y cmake libprotobuf-dev protobuf-compiler git \ - libgoogle-glog-dev libgflags-dev libatlas-dev libatlas3-base g++ m4 python-pip \ + libgoogle-glog-dev libgflags-dev libgtest-dev \ + libatlas-dev libatlas3-base g++ m4 python-pip \ python-protobuf python-numpy python-dev swig openssh-server \ wget unzip python-matplotlib tar xz-utils bzip2 gzip coreutils \ sed grep graphviz libjpeg-dev zlib1g-dev doxygen \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y +RUN cd /usr/src/gtest && cmake . && make && sudo cp *.a /usr/lib RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark From 174ca2723a5290f0130c616c2cfee248409de898 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sat, 10 Dec 2016 20:12:26 -0800 Subject: [PATCH 44/60] Remove sudo from Dockerfile.* --- paddle/scripts/docker/Dockerfile | 2 +- paddle/scripts/docker/Dockerfile.gpu | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/paddle/scripts/docker/Dockerfile b/paddle/scripts/docker/Dockerfile index 6243d1d188..bfe287f080 100644 --- a/paddle/scripts/docker/Dockerfile +++ b/paddle/scripts/docker/Dockerfile @@ -11,7 +11,7 @@ RUN apt-get update \ sed grep graphviz libjpeg-dev zlib1g-dev doxygen \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y -RUN cd /usr/src/gtest && cmake . && make && sudo cp *.a /usr/lib +RUN cd /usr/src/gtest && cmake . && make && cp *.a /usr/lib RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark diff --git a/paddle/scripts/docker/Dockerfile.gpu b/paddle/scripts/docker/Dockerfile.gpu index f9821f2760..26a88e414b 100644 --- a/paddle/scripts/docker/Dockerfile.gpu +++ b/paddle/scripts/docker/Dockerfile.gpu @@ -11,7 +11,7 @@ RUN apt-get update \ sed grep graphviz libjpeg-dev zlib1g-dev doxygen \ clang-3.8 llvm-3.8 libclang-3.8-dev \ && apt-get clean -y -RUN cd /usr/src/gtest && cmake . && make && sudo cp *.a /usr/lib +RUN cd /usr/src/gtest && cmake . && make && cp *.a /usr/lib RUN pip install -U BeautifulSoup docopt PyYAML pillow \ sphinx sphinx_rtd_theme breathe recommonmark From 469ea64d5780533d0892db9be90a39c403425c95 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Sun, 11 Dec 2016 12:26:22 +0800 Subject: [PATCH 45/60] Follow comments --- doc/howto/cmd_parameter/index_en.md | 2 +- doc/tutorials/rec/ml_dataset_en.md | 2 +- doc/tutorials/rec/ml_regression_en.rst | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/doc/howto/cmd_parameter/index_en.md b/doc/howto/cmd_parameter/index_en.md index a6c236db61..fb658f2aa5 100644 --- a/doc/howto/cmd_parameter/index_en.md +++ b/doc/howto/cmd_parameter/index_en.md @@ -1,5 +1,5 @@ ```eval_rst -.. _cmd_line_index: +.. _cmd_line_index_en: ``` # How to Set Command-line Parameters diff --git a/doc/tutorials/rec/ml_dataset_en.md b/doc/tutorials/rec/ml_dataset_en.md index 73879d6537..dc11a5e060 100644 --- a/doc/tutorials/rec/ml_dataset_en.md +++ b/doc/tutorials/rec/ml_dataset_en.md @@ -1,5 +1,5 @@ ```eval_rst -.. _demo_ml_dataset: +.. _demo_ml_dataset_en: ``` diff --git a/doc/tutorials/rec/ml_regression_en.rst b/doc/tutorials/rec/ml_regression_en.rst index b9c2f8cb59..ddc00dc706 100644 --- a/doc/tutorials/rec/ml_regression_en.rst +++ b/doc/tutorials/rec/ml_regression_en.rst @@ -16,7 +16,7 @@ Data Preparation ```````````````` Download and extract dataset '''''''''''''''''''''''''''' -We use :ref:`demo_ml_dataset` here. +We use :ref:`demo_ml_dataset_en` here. To download and unzip the dataset, simply run the following commands. .. code-block:: bash @@ -280,7 +280,7 @@ The run.sh is shown as follow: It just start a paddle training process, write the log to `log.txt`, then print it on screen. -Each command line argument in :code:`run.sh`, please refer to the :ref:`cmd_line_index` page. The short description of these arguments is shown as follow. +Each command line argument in :code:`run.sh`, please refer to the :ref:`cmd_line_index_en` page. The short description of these arguments is shown as follow. * config\: Tell paddle which file is neural network configuration. * save_dir\: Tell paddle save model into './output' From 62a7e5cf010a01601f8af4ddbe142a570752d4ed Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 11:16:13 -0800 Subject: [PATCH 46/60] Initialize Bazel WORKSPACE and a testing packages --- .gitignore | 3 +++ WORKSPACE | 8 ++++++++ third_party/protobuf_test/BUILD | 17 +++++++++++++++++ third_party/protobuf_test/README.md | 1 + third_party/protobuf_test/example.proto | 7 +++++++ third_party/protobuf_test/example_lib.cc | 6 ++++++ third_party/protobuf_test/example_lib.h | 7 +++++++ 7 files changed, 49 insertions(+) create mode 100644 WORKSPACE create mode 100644 third_party/protobuf_test/BUILD create mode 100644 third_party/protobuf_test/README.md create mode 100644 third_party/protobuf_test/example.proto create mode 100644 third_party/protobuf_test/example_lib.cc create mode 100644 third_party/protobuf_test/example_lib.h diff --git a/.gitignore b/.gitignore index 35bed0accd..1c9730a5ad 100644 --- a/.gitignore +++ b/.gitignore @@ -9,3 +9,6 @@ build/ .pydevproject Makefile .test_env/ + +*~ +bazel-* diff --git a/WORKSPACE b/WORKSPACE new file mode 100644 index 0000000000..06495690ab --- /dev/null +++ b/WORKSPACE @@ -0,0 +1,8 @@ +git_repository( + name = "org_pubref_rules_protobuf", + remote = "https://github.com/pubref/rules_protobuf", + tag = "v0.7.1", +) + +load("@org_pubref_rules_protobuf//cpp:rules.bzl", "cpp_proto_repositories") +cpp_proto_repositories() diff --git a/third_party/protobuf_test/BUILD b/third_party/protobuf_test/BUILD new file mode 100644 index 0000000000..7c5b1c6994 --- /dev/null +++ b/third_party/protobuf_test/BUILD @@ -0,0 +1,17 @@ +licenses(["notice"]) # Apache 2.0 + +load("@org_pubref_rules_protobuf//cpp:rules.bzl", "cpp_proto_library") + +cpp_proto_library( + name = "example_proto", + protos = [ + "example.proto" + ], +) + +cc_library( + name = "example_lib", + srcs = ["example_lib.cc"], + hdrs = ["example_lib.h"], + deps = [":example_proto"], +) diff --git a/third_party/protobuf_test/README.md b/third_party/protobuf_test/README.md new file mode 100644 index 0000000000..e8bdeee6fe --- /dev/null +++ b/third_party/protobuf_test/README.md @@ -0,0 +1 @@ +This package tests that Bazel can build protobuf related rules. diff --git a/third_party/protobuf_test/example.proto b/third_party/protobuf_test/example.proto new file mode 100644 index 0000000000..57c52d3521 --- /dev/null +++ b/third_party/protobuf_test/example.proto @@ -0,0 +1,7 @@ +syntax = "proto3"; + +package protos; + +message Greeting { + string name = 1; +} diff --git a/third_party/protobuf_test/example_lib.cc b/third_party/protobuf_test/example_lib.cc new file mode 100644 index 0000000000..8d55ed66de --- /dev/null +++ b/third_party/protobuf_test/example_lib.cc @@ -0,0 +1,6 @@ +#include "third_party/protobuf_test/example_lib.h" +#include + +std::string get_greet(const ::protos::Greeting& who) { + return "Hello " + who.name(); +} diff --git a/third_party/protobuf_test/example_lib.h b/third_party/protobuf_test/example_lib.h new file mode 100644 index 0000000000..eaf4dd4cea --- /dev/null +++ b/third_party/protobuf_test/example_lib.h @@ -0,0 +1,7 @@ +#pragma once + +#include "third_party/protobuf_test/example.pb.h" + +#include + +std::string get_greet(const ::protos::Greeting &who); From e74ae54a594ca2a68c936b570076de186bc87ae9 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 11:37:28 -0800 Subject: [PATCH 47/60] Update development document --- .../build_and_install/docker_install_en.rst | 66 +++++++++++++++++++ 1 file changed, 66 insertions(+) diff --git a/doc/getstarted/build_and_install/docker_install_en.rst b/doc/getstarted/build_and_install/docker_install_en.rst index 1ab6fc6a72..a3fa8bccbf 100644 --- a/doc/getstarted/build_and_install/docker_install_en.rst +++ b/doc/getstarted/build_and_install/docker_install_en.rst @@ -104,3 +104,69 @@ container: Then we can direct our Web browser to the HTML version of source code at http://localhost:8088/paddle/ + + +Development Using Docker +------------------------ + +Develpers can work on PaddlePaddle using Docker. This allows +developers to work on different platforms -- Linux, Mac OS X, and +Windows -- in a consistent way. + +The general development workflow with Docker and Bazel is as follows: + +1. Get the source code of Paddle: + + .. code-block:: bash + + git clone --recursive https://github.com/paddlepaddle/paddle + + +1. Build a development Docker image `paddle:dev` from the source code. + This image contains all the development tools and dependencies of + PaddlePaddle. + + + .. code-block:: bash + + cd paddle + docker build -t paddle:dev -f paddle/scripts/docker/Dockerfile . + + +1. Run the image as a container and mounting local source code + directory into the container. This allows us to change the code on + the host and build it within the container. + + .. code-block:: bash + + docker run \ + -d # run the container in background mode \ + --name paddle # we can run a nginx container to serve documents \ + -p 2022:22 # so we can SSH into this container \ + -v $PWD:/paddle # mount the source code \ + -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \ + paddle:dev + +1. SSH into the container: + + .. code-block:: bash + + ssh root@localhost -p 2022 + +1. We can edit the source code in the container or on this host. Then + we can build using cmake + + .. code-block:: bash + + cd /paddle # where paddle source code has been mounted into the container + mkdir -p build + cd build + cmake .. + make -j `nproc` + + or Bazel in the container: + + .. code-block:: bash + + cd /paddle + bazel build ... From 06d780070439761da6947a3f8b3c2342cf52f8cf Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 12:02:04 -0800 Subject: [PATCH 48/60] Correct indentation in .rst file --- .../build_and_install/docker_install_en.rst | 36 +++++++++---------- 1 file changed, 18 insertions(+), 18 deletions(-) diff --git a/doc/getstarted/build_and_install/docker_install_en.rst b/doc/getstarted/build_and_install/docker_install_en.rst index a3fa8bccbf..95b6954789 100644 --- a/doc/getstarted/build_and_install/docker_install_en.rst +++ b/doc/getstarted/build_and_install/docker_install_en.rst @@ -119,7 +119,7 @@ The general development workflow with Docker and Bazel is as follows: .. code-block:: bash - git clone --recursive https://github.com/paddlepaddle/paddle + git clone --recursive https://github.com/paddlepaddle/paddle 1. Build a development Docker image `paddle:dev` from the source code. @@ -129,8 +129,8 @@ The general development workflow with Docker and Bazel is as follows: .. code-block:: bash - cd paddle - docker build -t paddle:dev -f paddle/scripts/docker/Dockerfile . + cd paddle + docker build -t paddle:dev -f paddle/scripts/docker/Dockerfile . 1. Run the image as a container and mounting local source code @@ -139,34 +139,34 @@ The general development workflow with Docker and Bazel is as follows: .. code-block:: bash - docker run \ - -d # run the container in background mode \ - --name paddle # we can run a nginx container to serve documents \ - -p 2022:22 # so we can SSH into this container \ - -v $PWD:/paddle # mount the source code \ - -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \ - paddle:dev + docker run \ + -d # run the container in background mode \ + --name paddle # we can run a nginx container to serve documents \ + -p 2022:22 # so we can SSH into this container \ + -v $PWD:/paddle # mount the source code \ + -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \ + paddle:dev 1. SSH into the container: .. code-block:: bash - ssh root@localhost -p 2022 + ssh root@localhost -p 2022 1. We can edit the source code in the container or on this host. Then we can build using cmake .. code-block:: bash - cd /paddle # where paddle source code has been mounted into the container - mkdir -p build - cd build - cmake .. - make -j `nproc` + cd /paddle # where paddle source code has been mounted into the container + mkdir -p build + cd build + cmake .. + make -j `nproc` or Bazel in the container: .. code-block:: bash - cd /paddle - bazel build ... + cd /paddle + bazel build ... From 8013d17dfae690d6b0408fb6b0d8e4eceee30f2d Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 12:03:05 -0800 Subject: [PATCH 49/60] Correct enumeration in lists --- doc/getstarted/build_and_install/docker_install_en.rst | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/doc/getstarted/build_and_install/docker_install_en.rst b/doc/getstarted/build_and_install/docker_install_en.rst index 95b6954789..f6f1bbab42 100644 --- a/doc/getstarted/build_and_install/docker_install_en.rst +++ b/doc/getstarted/build_and_install/docker_install_en.rst @@ -122,7 +122,7 @@ The general development workflow with Docker and Bazel is as follows: git clone --recursive https://github.com/paddlepaddle/paddle -1. Build a development Docker image `paddle:dev` from the source code. +2. Build a development Docker image `paddle:dev` from the source code. This image contains all the development tools and dependencies of PaddlePaddle. @@ -133,7 +133,7 @@ The general development workflow with Docker and Bazel is as follows: docker build -t paddle:dev -f paddle/scripts/docker/Dockerfile . -1. Run the image as a container and mounting local source code +3. Run the image as a container and mounting local source code directory into the container. This allows us to change the code on the host and build it within the container. @@ -147,13 +147,13 @@ The general development workflow with Docker and Bazel is as follows: -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \ paddle:dev -1. SSH into the container: +4. SSH into the container: .. code-block:: bash ssh root@localhost -p 2022 -1. We can edit the source code in the container or on this host. Then +5. We can edit the source code in the container or on this host. Then we can build using cmake .. code-block:: bash From 9b94773a7f546f5d1a825c3a6f07420a5ced6404 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 12:39:02 -0800 Subject: [PATCH 50/60] Add gtest --- WORKSPACE | 14 ++++++++++++++ .../build_and_install/docker_install_en.rst | 5 +++-- third_party/gtest.BUILD | 14 ++++++++++++++ third_party/protobuf_test/BUILD | 11 +++++++++++ third_party/protobuf_test/example.proto | 2 +- third_party/protobuf_test/example_lib.cc | 9 +++++++-- third_party/protobuf_test/example_lib.h | 8 +++++++- third_party/protobuf_test/example_lib_test.cc | 15 +++++++++++++++ 8 files changed, 72 insertions(+), 6 deletions(-) create mode 100644 third_party/gtest.BUILD create mode 100644 third_party/protobuf_test/example_lib_test.cc diff --git a/WORKSPACE b/WORKSPACE index 06495690ab..38e1628d11 100644 --- a/WORKSPACE +++ b/WORKSPACE @@ -1,8 +1,22 @@ +# External dependency to grpc-enabled Google-styleprotobuf bulding +# rules. This method comes from +# https://github.com/pubref/rules_protobuf#usage. git_repository( name = "org_pubref_rules_protobuf", remote = "https://github.com/pubref/rules_protobuf", tag = "v0.7.1", ) +# External dependency to gtest 1.7.0. This method comes from +# https://www.bazel.io/versions/master/docs/tutorial/cpp.html. +new_http_archive( + name = "gtest", + url = "https://github.com/google/googletest/archive/release-1.7.0.zip", + sha256 = "b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0", + build_file = "third_party/gtest.BUILD", + strip_prefix = "googletest-release-1.7.0", +) + + load("@org_pubref_rules_protobuf//cpp:rules.bzl", "cpp_proto_repositories") cpp_proto_repositories() diff --git a/doc/getstarted/build_and_install/docker_install_en.rst b/doc/getstarted/build_and_install/docker_install_en.rst index f6f1bbab42..feb027ccbb 100644 --- a/doc/getstarted/build_and_install/docker_install_en.rst +++ b/doc/getstarted/build_and_install/docker_install_en.rst @@ -161,12 +161,13 @@ The general development workflow with Docker and Bazel is as follows: cd /paddle # where paddle source code has been mounted into the container mkdir -p build cd build - cmake .. + cmake -DWITH_TESTING=ON .. make -j `nproc` + CTEST_OUTPUT_ON_FAILURE=1 ctest or Bazel in the container: .. code-block:: bash cd /paddle - bazel build ... + bazel test ... diff --git a/third_party/gtest.BUILD b/third_party/gtest.BUILD new file mode 100644 index 0000000000..3e68a1d879 --- /dev/null +++ b/third_party/gtest.BUILD @@ -0,0 +1,14 @@ +cc_library( + name = "main", + srcs = glob( + ["src/*.cc"], + exclude = ["src/gtest-all.cc"] + ), + hdrs = glob([ + "include/**/*.h", + "src/*.h" + ]), + copts = ["-Iexternal/gtest/include"], + linkopts = ["-pthread"], + visibility = ["//visibility:public"], +) diff --git a/third_party/protobuf_test/BUILD b/third_party/protobuf_test/BUILD index 7c5b1c6994..6208b87082 100644 --- a/third_party/protobuf_test/BUILD +++ b/third_party/protobuf_test/BUILD @@ -7,6 +7,7 @@ cpp_proto_library( protos = [ "example.proto" ], + with_grpc = True, ) cc_library( @@ -15,3 +16,13 @@ cc_library( hdrs = ["example_lib.h"], deps = [":example_proto"], ) + +cc_test( + name = "example_lib_test", + srcs = ["example_lib_test.cc"], + copts = ["-Iexternal/gtest/include"], + deps =[ + "@gtest//:main", + ":example_lib", + ], +) diff --git a/third_party/protobuf_test/example.proto b/third_party/protobuf_test/example.proto index 57c52d3521..6a7eada9c1 100644 --- a/third_party/protobuf_test/example.proto +++ b/third_party/protobuf_test/example.proto @@ -1,6 +1,6 @@ syntax = "proto3"; -package protos; +package third_party.protobuf_test; message Greeting { string name = 1; diff --git a/third_party/protobuf_test/example_lib.cc b/third_party/protobuf_test/example_lib.cc index 8d55ed66de..56341a0124 100644 --- a/third_party/protobuf_test/example_lib.cc +++ b/third_party/protobuf_test/example_lib.cc @@ -1,6 +1,11 @@ #include "third_party/protobuf_test/example_lib.h" -#include -std::string get_greet(const ::protos::Greeting& who) { +namespace third_party { +namespace protobuf_test { + +std::string get_greet(const Greeting& who) { return "Hello " + who.name(); } + +} // namespace protobuf_test +} // namespace thrid_party diff --git a/third_party/protobuf_test/example_lib.h b/third_party/protobuf_test/example_lib.h index eaf4dd4cea..516326e812 100644 --- a/third_party/protobuf_test/example_lib.h +++ b/third_party/protobuf_test/example_lib.h @@ -4,4 +4,10 @@ #include -std::string get_greet(const ::protos::Greeting &who); +namespace third_party { +namespace protobuf_test { + +std::string get_greet(const Greeting &who); + +} // namespace protobuf_test +} // namespace third_party diff --git a/third_party/protobuf_test/example_lib_test.cc b/third_party/protobuf_test/example_lib_test.cc new file mode 100644 index 0000000000..6229f56e60 --- /dev/null +++ b/third_party/protobuf_test/example_lib_test.cc @@ -0,0 +1,15 @@ +#include "third_party/protobuf_test/example_lib.h" + +#include "gtest/gtest.h" + +namespace third_party { +namespace protobuf_test { + +TEST(ProtobufTest, GetGreet) { + Greeting g; + g.set_name("Paddle"); + EXPECT_EQ("Hello Paddle", get_greet(g)); +} + +} // namespace protobuf_test +} // namespace third_party From dab81d4d030f08466448312a7697b7bf94806927 Mon Sep 17 00:00:00 2001 From: dangqingqing Date: Mon, 12 Dec 2016 11:28:05 +0800 Subject: [PATCH 51/60] fix dead links in quick start after reorganzing documentaion directory. --- doc/tutorials/quick_start/index_en.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/doc/tutorials/quick_start/index_en.md b/doc/tutorials/quick_start/index_en.md index ec548b5393..2ade741799 100644 --- a/doc/tutorials/quick_start/index_en.md +++ b/doc/tutorials/quick_start/index_en.md @@ -12,7 +12,7 @@ This tutorial will teach the basics of deep learning (DL), including how to impl To get started, please install PaddlePaddle on your computer. Throughout this tutorial, you will learn by implementing different DL models for text classification. -To install PaddlePaddle, please follow the instructions here: Build and Install. +To install PaddlePaddle, please follow the instructions here: Build and Install. ## Overview For the first step, you will use PaddlePaddle to build a **text classification** system. For example, suppose you run an e-commence website, and you want to analyze the sentiment of user reviews to evaluate product quality. @@ -156,14 +156,14 @@ define_py_data_sources2(train_list='data/train.list', obj="process", args={"dictionary": word_dict}) ``` -You can refer to the following link for more detailed examples and data formats: PyDataProvider2. +You can refer to the following link for more detailed examples and data formats: PyDataProvider2. ## Network Architecture You will describe four kinds of network architectures in this section.
![](./PipelineNetwork_en.jpg)
First, you will build a logistic regression model. Later, you will also get chance to build other more powerful network architectures. -For more detailed documentation, you could refer to: Layer documentation。All configuration files are in `demo/quick_start` directory. +For more detailed documentation, you could refer to: layer documentation. All configuration files are in `demo/quick_start` directory. ### Logistic Regression The architecture is illustrated in the following picture: @@ -366,7 +366,7 @@ You can use single layer LSTM model with Dropout for our text classification pro
## Optimization Algorithm -Optimization algorithms include Momentum, RMSProp, AdaDelta, AdaGrad, Adam, and Adamax. You can use Adam optimization method here, with L2 regularization and gradient clipping, because Adam has been proved to work very well for training recurrent neural network. +Optimization algorithms include Momentum, RMSProp, AdaDelta, AdaGrad, Adam, and Adamax. You can use Adam optimization method here, with L2 regularization and gradient clipping, because Adam has been proved to work very well for training recurrent neural network. ```python settings(batch_size=128, @@ -391,7 +391,7 @@ paddle train \ --use_gpu=false ``` -If you want to install the remote training platform, which enables distributed training on clusters, follow the instructions here: Platform documentation. We do not provide examples on how to train on clusters. Please refer to other demos or platform training documentation for mode details on training on clusters. +If you want to install the remote training platform, which enables distributed training on clusters, follow the instructions here: Platform documentation. We do not provide examples on how to train on clusters. Please refer to other demos or platform training documentation for mode details on training on clusters. ## Inference You can use the trained model to perform prediction on the dataset with no labels. You can also evaluate the model on dataset with labels to obtain its test accuracy.
![](./PipelineTest_en.png)
@@ -406,7 +406,7 @@ paddle train \ --init_model_path=./output/pass-0000x ``` -We will give an example of performing prediction using Recurrent model on a dataset with no labels. You can refer to: Python Prediction API tutorial，or other demo for the prediction process using Python. You can also use the following script for inference or evaluation. +We will give an example of performing prediction using Recurrent model on a dataset with no labels. You can refer to Python Prediction API tutorial，or other demo for the prediction process using Python. You can also use the following script for inference or evaluation. inference script (predict.sh)： @@ -508,7 +508,7 @@ The scripts of data downloading, network configurations, and training scrips are * \--config_args：Other configuration arguments. * \--init_model_path：The path of the initial model parameter. -By default, the trainer will save model every pass. You can also specify `saving_period_by_batches` to set the frequency of batch saving. You can use `show_parameter_stats_period` to print the statistics of the parameters, which are very useful for tuning parameters. Other command line arguments can be found in command line argument documentation。 +By default, the trainer will save model every pass. You can also specify `saving_period_by_batches` to set the frequency of batch saving. You can use `show_parameter_stats_period` to print the statistics of the parameters, which are very useful for tuning parameters. Other command line arguments can be found in command line argument documentation。 ### Log From f722ee21acdc190f456a64173647c38bffabbdb8 Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Mon, 12 Dec 2016 04:19:34 +0000 Subject: [PATCH 52/60] Do not use https://github.com/pubref/rules_protobuf, but use https://github.com/google/protobuf/ as external dependency for protobuf --- .pre-commit-config.yaml | 3 ++- WORKSPACE | 17 ++++++----------- third_party/protobuf_test/BUILD | 11 +++++------ 3 files changed, 13 insertions(+), 18 deletions(-) diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml index 942669c41f..641ac583b5 100644 --- a/.pre-commit-config.yaml +++ b/.pre-commit-config.yaml @@ -6,7 +6,8 @@ - repo: https://github.com/reyoung/mirrors-yapf.git sha: v0.13.2 hooks: - - id: yapf + - id: yapf + files: (.*\.(py|bzl)|BUILD|.*\.BUILD)$ # Bazel BUILD files follow Python syntax. - repo: https://github.com/pre-commit/pre-commit-hooks sha: 7539d8bd1a00a3c1bfd34cdb606d3a6372e83469 hooks: diff --git a/WORKSPACE b/WORKSPACE index 38e1628d11..d6ae2af8eb 100644 --- a/WORKSPACE +++ b/WORKSPACE @@ -1,10 +1,9 @@ -# External dependency to grpc-enabled Google-styleprotobuf bulding -# rules. This method comes from -# https://github.com/pubref/rules_protobuf#usage. -git_repository( - name = "org_pubref_rules_protobuf", - remote = "https://github.com/pubref/rules_protobuf", - tag = "v0.7.1", +# External dependency to Google protobuf. +http_archive( + name = "protobuf", + url = "http://github.com/google/protobuf/archive/v3.1.0.tar.gz", + sha256 = "0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7", + strip_prefix = "protobuf-3.1.0", ) # External dependency to gtest 1.7.0. This method comes from @@ -16,7 +15,3 @@ new_http_archive( build_file = "third_party/gtest.BUILD", strip_prefix = "googletest-release-1.7.0", ) - - -load("@org_pubref_rules_protobuf//cpp:rules.bzl", "cpp_proto_repositories") -cpp_proto_repositories() diff --git a/third_party/protobuf_test/BUILD b/third_party/protobuf_test/BUILD index 6208b87082..46f769da5f 100644 --- a/third_party/protobuf_test/BUILD +++ b/third_party/protobuf_test/BUILD @@ -1,13 +1,12 @@ licenses(["notice"]) # Apache 2.0 -load("@org_pubref_rules_protobuf//cpp:rules.bzl", "cpp_proto_library") +load("@protobuf//:protobuf.bzl", "cc_proto_library") -cpp_proto_library( +cc_proto_library( name = "example_proto", - protos = [ - "example.proto" - ], - with_grpc = True, + srcs = ["example.proto"], + protoc = "@protobuf//:protoc", + default_runtime = "@protobuf//:protobuf", ) cc_library( From 7e7b2bb7138a481c2004ae9e27ce15cc5e3c2adc Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 20:26:22 -0800 Subject: [PATCH 53/60] Make Git pre-commit hook auto-format WORKSPACE as well. --- .pre-commit-config.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml index 641ac583b5..b9902a863d 100644 --- a/.pre-commit-config.yaml +++ b/.pre-commit-config.yaml @@ -7,7 +7,7 @@ sha: v0.13.2 hooks: - id: yapf - files: (.*\.(py|bzl)|BUILD|.*\.BUILD)$ # Bazel BUILD files follow Python syntax. + files: (.*\.(py|bzl)|BUILD|.*\.BUILD|WORKSPACE)$ # Bazel BUILD files follow Python syntax. - repo: https://github.com/pre-commit/pre-commit-hooks sha: 7539d8bd1a00a3c1bfd34cdb606d3a6372e83469 hooks: From 2a5a8e79e5e01c99670cd647d7083ea00cbaf0ec Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 14:15:46 +0800 Subject: [PATCH 54/60] Fix Travis-CI when using rebase. * The $TRAVIS_COMMIT_RANGE could be not existed, because the git history can be changed and rebased. Here we alway run Travis-CI when `git diff` cannot return the modified files. * Also do not push docs update if is not PaddlePaddle/Paddle. --- .travis.yml | 11 +++++++---- paddle/scripts/travis/docs.sh | 31 +++++++++++++++++-------------- 2 files changed, 24 insertions(+), 18 deletions(-) diff --git a/.travis.yml b/.travis.yml index 6215060e33..145d22c308 100644 --- a/.travis.yml +++ b/.travis.yml @@ -42,10 +42,13 @@ addons: before_install: - | if [ ${JOB} == "BUILD_AND_TEST" ]; then - if ! git diff --name-only $TRAVIS_COMMIT_RANGE | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)' - then - echo "Only markdown docs were updated, stopping build process." - exit + local change_list=`git diff --name-only $TRAVIS_COMMIT_RANGE` + if [ $? -eq 0 ]; then # if git diff return no zero, then rerun unit test. + if ! echo ${change_list} | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)' + then + echo "Only markdown docs were updated, stopping build process." + exit + fi fi fi - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then sudo paddle/scripts/travis/before_install.linux.sh; fi diff --git a/paddle/scripts/travis/docs.sh b/paddle/scripts/travis/docs.sh index c2a4809d75..0bbb76a8a3 100755 --- a/paddle/scripts/travis/docs.sh +++ b/paddle/scripts/travis/docs.sh @@ -47,17 +47,20 @@ if [ $? -eq 0 ]; then fi set -e -# Commit -git add . -git config user.name "Travis CI" -git config user.email "paddle-dev@baidu.com" -git commit -m "Deploy to GitHub Pages: ${SHA}" - -# Set ssh private key -openssl aes-256-cbc -K $SSL_KEY -iv $SSL_IV -in ../../paddle/scripts/travis/deploy_key.enc -out deploy_key -d -chmod 600 deploy_key -eval `ssh-agent -s` -ssh-add deploy_key - -# Push -git push $SSH_REPO $TARGET_BRANCH +if [ -n $SSL_KEY ]; then # Only push updated docs for github.com/PaddlePaddle/Paddle. + # Commit + git add . + git config user.name "Travis CI" + git config user.email "paddle-dev@baidu.com" + git commit -m "Deploy to GitHub Pages: ${SHA}" + + # Set ssh private key + openssl aes-256-cbc -K $SSL_KEY -iv $SSL_IV -in ../../paddle/scripts/travis/deploy_key.enc -out deploy_key -d + chmod 600 deploy_key + eval `ssh-agent -s` + ssh-add deploy_key + + # Push + git push $SSH_REPO $TARGET_BRANCH + +fi From 068bfbb817611c856acd8c535de2b33a6126786c Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 14:48:32 +0800 Subject: [PATCH 55/60] All file pass pre-commit hook --- .travis.yml | 9 +- benchmark/tensorflow/rnn/run_multi.sh | 1 - demo/gan/README.md | 2 +- demo/gan/data/download_cifar.sh | 1 - demo/gan/data/get_mnist_data.sh | 2 - demo/gan/gan_conf.py | 147 +++---- demo/gan/gan_conf_image.py | 358 ++++++++++-------- demo/gan/gan_trainer.py | 134 ++++--- .../quick_start/trainer_config.resnet-lstm.py | 27 +- .../data/extract_dict_feature.py | 5 +- .../data/extract_pairs.py | 8 +- demo/semantic_role_labeling/dataprovider.py | 19 +- demo/semantic_role_labeling/db_lstm.py | 174 ++++----- demo/semantic_role_labeling/predict.py | 44 +-- .../k8s/distributed_training_on_kubernetes.md | 2 +- doc_cn/cluster/k8s/job.yaml | 2 +- doc_cn/cluster/k8s/start_paddle.py | 5 +- doc_cn/demo/sentiment_analysis/index.rst | 14 +- doc_theme/static/js/paddle_doc_init.js | 2 +- paddle/api/GradientMachine.cpp | 2 +- paddle/api/Internal.h | 9 +- paddle/api/Matrix.cpp | 8 +- paddle/api/PaddleAPI.h | 2 +- paddle/api/Parameter.cpp | 2 +- paddle/api/ParameterOptimizer.cpp | 23 +- paddle/api/SequenceGenerator.cpp | 8 +- paddle/api/Trainer.cpp | 4 +- paddle/api/Util.cpp | 10 +- paddle/api/Vector.cpp | 2 +- paddle/api/test/testMatrix.py | 5 +- paddle/api/test/testVector.py | 18 +- paddle/cuda/include/hl_base.h | 4 +- paddle/cuda/include/hl_dso_loader.h | 2 +- paddle/cuda/include/hl_gpu.h | 18 +- paddle/cuda/src/hl_cuda_cublas.cc | 4 +- paddle/cuda/src/hl_cuda_cudnn.cc | 6 +- paddle/cuda/src/hl_cudart_wrap.cc | 2 +- paddle/cuda/src/hl_time.cc | 4 +- paddle/cuda/src/hl_warpctc_wrap.cc | 2 +- .../activations/ActivationFunction.cpp | 6 +- paddle/gserver/dataproviders/DataProvider.cpp | 8 +- paddle/gserver/dataproviders/DataProvider.h | 28 +- .../dataproviders/MultiDataProvider.cpp | 4 +- .../dataproviders/ProtoDataProvider.cpp | 42 +- .../gserver/dataproviders/ProtoDataProvider.h | 2 +- paddle/gserver/dataproviders/ProtoReader.h | 4 +- .../gserver/dataproviders/PyDataProvider.cpp | 40 +- .../gserver/dataproviders/PyDataProvider2.cpp | 23 +- paddle/gserver/evaluators/Evaluator.cpp | 8 +- paddle/gserver/evaluators/Evaluator.h | 6 +- .../gradientmachines/GradientMachine.cpp | 12 +- .../gradientmachines/GradientMachine.h | 10 +- .../gradientmachines/MultiGradientMachine.h | 4 +- .../gserver/gradientmachines/MultiNetwork.cpp | 2 +- .../gradientmachines/NeuralNetwork.cpp | 8 +- .../gserver/gradientmachines/NeuralNetwork.h | 25 +- .../gradientmachines/ParallelNeuralNetwork.h | 15 +- .../RecurrentGradientMachine.cpp | 49 +-- .../RecurrentGradientMachine.h | 2 +- paddle/gserver/layers/BatchNormBaseLayer.cpp | 4 +- paddle/gserver/layers/BatchNormBaseLayer.h | 2 +- .../gserver/layers/BatchNormalizationLayer.h | 2 +- paddle/gserver/layers/ConcatenateLayer.cpp | 2 +- paddle/gserver/layers/ContextProjection.cpp | 2 +- paddle/gserver/layers/ConvBaseLayer.cpp | 2 +- paddle/gserver/layers/ConvOperator.cpp | 4 +- paddle/gserver/layers/ConvProjection.cpp | 2 +- paddle/gserver/layers/ConvShiftLayer.cpp | 2 +- .../gserver/layers/ConvexCombinationLayer.cpp | 2 +- paddle/gserver/layers/CosSimVecMatLayer.cpp | 2 +- paddle/gserver/layers/CostLayer.cpp | 6 +- paddle/gserver/layers/CudnnBatchNormLayer.cpp | 4 +- paddle/gserver/layers/CudnnBatchNormLayer.h | 4 +- paddle/gserver/layers/CudnnConvLayer.cpp | 2 +- paddle/gserver/layers/CudnnConvLayer.h | 4 +- paddle/gserver/layers/CudnnPoolLayer.cpp | 4 +- paddle/gserver/layers/EosIdCheckLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvBaseLayer.h | 2 +- paddle/gserver/layers/ExpandConvLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvLayer.h | 2 +- .../gserver/layers/ExpandConvTransLayer.cpp | 2 +- paddle/gserver/layers/ExpandConvTransLayer.h | 2 +- paddle/gserver/layers/FullyConnectedLayer.cpp | 6 +- paddle/gserver/layers/GatedRecurrentLayer.cpp | 7 +- paddle/gserver/layers/GatedRecurrentLayer.h | 4 +- paddle/gserver/layers/GruCompute.cpp | 2 +- paddle/gserver/layers/GruCompute.h | 2 +- paddle/gserver/layers/GruStepLayer.cpp | 2 +- paddle/gserver/layers/IdentityProjection.cpp | 2 +- paddle/gserver/layers/InterpolationLayer.cpp | 2 +- paddle/gserver/layers/Layer.cpp | 6 +- paddle/gserver/layers/Layer.h | 10 +- paddle/gserver/layers/LinearChainCRF.cpp | 2 +- paddle/gserver/layers/LinearChainCTC.cpp | 2 +- paddle/gserver/layers/LstmCompute.cpp | 4 +- paddle/gserver/layers/LstmCompute.h | 2 +- paddle/gserver/layers/LstmLayer.cpp | 2 +- paddle/gserver/layers/LstmLayer.h | 6 +- paddle/gserver/layers/MDLstmLayer.cpp | 4 +- paddle/gserver/layers/MaxOutLayer.cpp | 2 +- paddle/gserver/layers/MixedLayer.cpp | 2 +- paddle/gserver/layers/MixedLayer.h | 2 +- paddle/gserver/layers/MultiplexLayer.cpp | 2 +- paddle/gserver/layers/NormLayer.cpp | 2 +- paddle/gserver/layers/NormLayer.h | 2 +- paddle/gserver/layers/NormProjectionLayer.cpp | 2 +- paddle/gserver/layers/NormProjectionLayer.h | 2 +- paddle/gserver/layers/Operator.h | 4 +- paddle/gserver/layers/OuterProdLayer.cpp | 2 +- paddle/gserver/layers/PoolLayer.cpp | 2 +- paddle/gserver/layers/PoolLayer.h | 4 +- paddle/gserver/layers/PoolProjectionLayer.cpp | 2 +- paddle/gserver/layers/PowerLayer.cpp | 2 +- paddle/gserver/layers/RecurrentLayer.cpp | 2 +- paddle/gserver/layers/RecurrentLayerGroup.cpp | 2 +- paddle/gserver/layers/ResizeLayer.cpp | 2 +- paddle/gserver/layers/ScalingLayer.cpp | 2 +- .../layers/SelectiveFullyConnectedLayer.cpp | 6 +- paddle/gserver/layers/SequenceConcatLayer.cpp | 2 +- paddle/gserver/layers/SequencePoolLayer.cpp | 2 +- .../gserver/layers/SequenceReshapeLayer.cpp | 2 +- paddle/gserver/layers/SequenceToBatch.cpp | 6 +- paddle/gserver/layers/SequenceToBatch.h | 2 +- paddle/gserver/layers/SlopeInterceptLayer.cpp | 2 +- paddle/gserver/layers/SubSequenceLayer.cpp | 2 +- paddle/gserver/layers/SumToOneNormLayer.cpp | 2 +- paddle/gserver/layers/TransLayer.cpp | 2 +- paddle/gserver/layers/TransLayer.h | 2 +- .../layers/TransposedFullMatrixProjection.cpp | 2 +- paddle/gserver/layers/ValidationLayer.cpp | 4 +- paddle/gserver/layers/ValidationLayer.h | 2 +- paddle/gserver/tests/LayerGradUtil.h | 4 +- paddle/gserver/tests/TestUtil.cpp | 10 +- paddle/gserver/tests/test_ActivationGrad.cpp | 6 +- paddle/gserver/tests/test_BatchNorm.cpp | 157 ++++---- paddle/gserver/tests/test_ConvTrans.cpp | 10 +- paddle/gserver/tests/test_ConvUnify.cpp | 246 ++++++------ paddle/gserver/tests/test_Evaluator.cpp | 2 +- paddle/gserver/tests/test_LayerGrad.cpp | 2 +- .../gserver/tests/test_MultinomialSampler.cpp | 2 +- paddle/gserver/tests/test_NetworkCompare.cpp | 6 +- .../gserver/tests/test_ProtoDataProvider.cpp | 2 +- paddle/gserver/tests/test_RecurrentLayer.cpp | 6 +- .../gserver/tests/test_SelectiveFCLayer.cpp | 10 +- paddle/gserver/tests/test_WarpCTCLayer.cpp | 6 +- paddle/math/Allocator.h | 2 +- paddle/math/BaseMatrix.h | 4 +- paddle/math/CpuSparseMatrix.cpp | 10 +- paddle/math/MathFunctions.cpp | 2 +- paddle/math/MathUtils.cpp | 2 +- paddle/math/Matrix.h | 6 +- paddle/math/MatrixBitCode.cpp | 4 +- paddle/math/MemoryHandle.cpp | 2 +- paddle/math/PoolAllocator.h | 4 +- paddle/math/SparseMatrix.cpp | 14 +- paddle/math/SparseMatrix.h | 2 +- paddle/math/SparseRowMatrix.h | 4 +- paddle/math/Storage.cpp | 4 +- paddle/math/Storage.h | 2 +- paddle/math/TensorEvaluate.h | 2 +- paddle/math/TensorExpression.h | 6 +- paddle/math/TrainingAlgorithmOp.h | 2 +- paddle/math/Vector.cpp | 13 +- paddle/math/Vector.h | 6 +- paddle/math/tests/OriginalOptimizerApi.h | 2 +- paddle/math/tests/TestUtils.h | 2 +- paddle/math/tests/test_Allocator.cpp | 4 +- paddle/math/tests/test_BaseMatrix.cpp | 2 +- paddle/math/tests/test_CpuGpuVector.cpp | 4 +- paddle/math/tests/test_ExecViaCpu.cpp | 4 +- paddle/math/tests/test_GpuProfiler.cpp | 64 +++- paddle/math/tests/test_SIMDFunctions.cpp | 4 +- paddle/math/tests/test_TrainingAlgorithm.cpp | 6 +- paddle/math/tests/test_batchTranspose.cpp | 2 +- paddle/math/tests/test_matrixCompare.cpp | 8 +- paddle/math/tests/test_perturbation.cpp | 4 +- .../math/tests/test_sparseMatrixCompare.cpp | 4 +- paddle/parameter/Argument.cpp | 9 +- paddle/parameter/Argument.h | 2 +- paddle/parameter/FirstOrderOptimizer.cpp | 6 +- paddle/parameter/ParallelParameter.cpp | 2 +- paddle/parameter/ParallelParameter.h | 10 +- paddle/parameter/Parameter.cpp | 10 +- paddle/parameter/Parameter.h | 10 +- paddle/parameter/ParameterUpdateFunctions.h | 2 +- paddle/parameter/ParameterUpdaterBase.cpp | 4 +- paddle/parameter/ParameterUpdaterHook.cpp | 9 +- paddle/parameter/Regularizer.cpp | 4 +- paddle/parameter/Weight.cpp | 2 +- paddle/parameter/tests/test_common.cpp | 4 +- paddle/pserver/BaseClient.cpp | 6 +- paddle/pserver/BaseClient.h | 4 +- paddle/pserver/LightNetwork.cpp | 12 +- paddle/pserver/LightNetwork.h | 2 +- paddle/pserver/ParameterClient2.cpp | 4 +- paddle/pserver/ParameterClient2.h | 12 +- paddle/pserver/ParameterServer2.cpp | 6 +- paddle/pserver/ParameterServer2.h | 12 +- paddle/pserver/ParameterServer2Main.cpp | 6 +- paddle/pserver/ProtoServer.h | 3 +- paddle/pserver/SocketChannel.cpp | 6 +- paddle/pserver/SparseParameterDistribution.h | 2 +- paddle/pserver/test/SocketTest.cpp | 6 +- paddle/pserver/test/test_ParameterServer2.cpp | 2 +- paddle/pserver/test/test_ProtoServer.cpp | 4 +- paddle/py_paddle/util.py | 10 +- paddle/scripts/travis/main.sh | 2 + paddle/scripts/travis/precommit.sh | 6 + paddle/trainer/MergeModel.cpp | 4 +- paddle/trainer/ParamUtil.cpp | 8 +- paddle/trainer/ParamUtil.h | 6 +- paddle/trainer/ParameterUpdater.h | 2 +- paddle/trainer/RemoteParameterUpdater.cpp | 2 +- paddle/trainer/RemoteParameterUpdater.h | 6 +- paddle/trainer/Tester.h | 6 +- paddle/trainer/TesterConfig.h | 4 +- paddle/trainer/ThreadParameterUpdater.h | 2 +- paddle/trainer/Trainer.h | 10 +- paddle/trainer/TrainerConfigHelper.h | 2 +- paddle/trainer/TrainerInternal.cpp | 12 +- paddle/trainer/TrainerInternal.h | 8 +- paddle/trainer/TrainerInternalConfig.h | 4 +- paddle/trainer/TrainerMain.cpp | 4 +- paddle/trainer/tests/picojson.h | 2 +- paddle/trainer/tests/test_Compare.cpp | 2 +- paddle/trainer/tests/test_CompareTwoNets.cpp | 4 +- paddle/trainer/tests/test_CompareTwoOpts.cpp | 4 +- .../tests/test_PyDataProviderWrapper.cpp | 8 +- paddle/trainer/tests/test_TrainerOnePass.cpp | 2 +- .../test_recurrent_machine_generation.cpp | 2 +- paddle/utils/BarrierStat.cpp | 8 +- paddle/utils/BarrierStat.h | 10 +- paddle/utils/CommandLineParser.cpp | 27 +- paddle/utils/CommandLineParser.h | 4 +- paddle/utils/CpuId.cpp | 46 +-- paddle/utils/CpuId.h | 84 ++-- paddle/utils/CustomStackTrace.cpp | 2 +- paddle/utils/CustomStackTrace.h | 5 +- paddle/utils/Logging.cpp | 8 +- paddle/utils/Logging.h | 2 +- paddle/utils/PythonUtil.cpp | 2 +- paddle/utils/PythonUtil.h | 4 +- paddle/utils/Queue.h | 6 +- paddle/utils/Stat.cpp | 9 +- paddle/utils/StringUtil.h | 2 +- paddle/utils/Thread.h | 4 +- paddle/utils/ThreadLocal.cpp | 2 +- paddle/utils/ThreadLocal.h | 4 +- paddle/utils/Util.cpp | 40 +- paddle/utils/Util.h | 16 +- paddle/utils/Version.cpp | 7 +- paddle/utils/Version.h | 2 +- paddle/utils/arch/osx/Locks.cpp | 4 +- paddle/utils/tests/test_CommandLineParser.cpp | 2 +- paddle/utils/tests/test_CustomStackTrace.cpp | 4 +- .../tests/test_CustomStackTracePrint.cpp | 2 +- paddle/utils/tests/test_Logging.cpp | 4 +- paddle/utils/tests/test_SIMDFlags.cpp | 41 +- paddle/utils/tests/test_SpinLock.cpp | 4 +- paddle/utils/tests/test_Thread.cpp | 22 +- paddle/utils/tests/test_ThreadBarrier.cpp | 4 +- 261 files changed, 1524 insertions(+), 1404 deletions(-) create mode 100755 paddle/scripts/travis/precommit.sh diff --git a/.travis.yml b/.travis.yml index 6215060e33..9fc518b0cc 100644 --- a/.travis.yml +++ b/.travis.yml @@ -8,10 +8,13 @@ os: env: - JOB=DOCS - JOB=BUILD_AND_TEST + - JOB=PRE_COMMIT matrix: exclude: - os: osx - env: JOB=DOCS # Only generate documentation in linux + env: JOB=DOCS # Only generate documentation in linux. + - os: osx + env: JOB=PRE_COMMIT # Only check pre-commit hook in linux addons: apt: @@ -39,6 +42,7 @@ addons: - lcov - graphviz - swig + - clang-format-3.8 before_install: - | if [ ${JOB} == "BUILD_AND_TEST" ]; then @@ -50,7 +54,8 @@ before_install: fi - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then sudo paddle/scripts/travis/before_install.linux.sh; fi - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then paddle/scripts/travis/before_install.osx.sh; fi - - pip install wheel protobuf sphinx breathe recommonmark virtualenv numpy sphinx_rtd_theme + - if [[ "$JOB" == "PRE_COMMIT" ]]; then sudo ln -s /usr/bin/clang-format-3.8 /usr/bin/clang-format; fi + - pip install wheel protobuf sphinx breathe recommonmark virtualenv numpy sphinx_rtd_theme pre-commit script: - paddle/scripts/travis/main.sh notifications: diff --git a/benchmark/tensorflow/rnn/run_multi.sh b/benchmark/tensorflow/rnn/run_multi.sh index f7f52e01e3..c2d7dd597e 100755 --- a/benchmark/tensorflow/rnn/run_multi.sh +++ b/benchmark/tensorflow/rnn/run_multi.sh @@ -25,4 +25,3 @@ test 4 2 256 512 test 4 2 512 128 test 4 2 512 256 test 4 2 512 512 - diff --git a/demo/gan/README.md b/demo/gan/README.md index fdc970a07b..1908b534b0 100644 --- a/demo/gan/README.md +++ b/demo/gan/README.md @@ -10,4 +10,4 @@ Then you can run the command below. The flag -d specifies the training data (cif $python gan_trainer.py -d cifar --use_gpu 1 The generated images will be stored in ./cifar_samples/ -The corresponding models will be stored in ./cifar_params/ \ No newline at end of file +The corresponding models will be stored in ./cifar_params/ diff --git a/demo/gan/data/download_cifar.sh b/demo/gan/data/download_cifar.sh index 32e73b3d8e..ae24ef2b7f 100755 --- a/demo/gan/data/download_cifar.sh +++ b/demo/gan/data/download_cifar.sh @@ -15,4 +15,3 @@ set -e wget https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz tar zxf cifar-10-python.tar.gz rm cifar-10-python.tar.gz - diff --git a/demo/gan/data/get_mnist_data.sh b/demo/gan/data/get_mnist_data.sh index d21bf70671..a77c81bf5a 100644 --- a/demo/gan/data/get_mnist_data.sh +++ b/demo/gan/data/get_mnist_data.sh @@ -15,5 +15,3 @@ do gunzip ${fname}.gz fi done - - diff --git a/demo/gan/gan_conf.py b/demo/gan/gan_conf.py index 58ba9dde58..86ac2dffe5 100644 --- a/demo/gan/gan_conf.py +++ b/demo/gan/gan_conf.py @@ -14,10 +14,9 @@ from paddle.trainer_config_helpers import * mode = get_config_arg("mode", str, "generator") -assert mode in set(["generator", - "discriminator", - "generator_training", - "discriminator_training"]) +assert mode in set([ + "generator", "discriminator", "generator_training", "discriminator_training" +]) is_generator_training = mode == "generator_training" is_discriminator_training = mode == "discriminator_training" @@ -38,8 +37,8 @@ sample_dim = 2 settings( batch_size=128, learning_rate=1e-4, - learning_method=AdamOptimizer(beta1=0.5) -) + learning_method=AdamOptimizer(beta1=0.5)) + def discriminator(sample): """ @@ -50,70 +49,87 @@ def discriminator(sample): of the sample is from real data. """ param_attr = ParamAttr(is_static=is_generator_training) - bias_attr = ParamAttr(is_static=is_generator_training, - initial_mean=1.0, - initial_std=0) - - hidden = fc_layer(input=sample, name="dis_hidden", size=hidden_dim, - bias_attr=bias_attr, - param_attr=param_attr, - act=ReluActivation()) - - hidden2 = fc_layer(input=hidden, name="dis_hidden2", size=hidden_dim, - bias_attr=bias_attr, - param_attr=param_attr, - act=LinearActivation()) - - hidden_bn = batch_norm_layer(hidden2, - act=ReluActivation(), - name="dis_hidden_bn", - bias_attr=bias_attr, - param_attr=ParamAttr(is_static=is_generator_training, - initial_mean=1.0, - initial_std=0.02), - use_global_stats=False) - - return fc_layer(input=hidden_bn, name="dis_prob", size=2, - bias_attr=bias_attr, - param_attr=param_attr, - act=SoftmaxActivation()) + bias_attr = ParamAttr( + is_static=is_generator_training, initial_mean=1.0, initial_std=0) + + hidden = fc_layer( + input=sample, + name="dis_hidden", + size=hidden_dim, + bias_attr=bias_attr, + param_attr=param_attr, + act=ReluActivation()) + + hidden2 = fc_layer( + input=hidden, + name="dis_hidden2", + size=hidden_dim, + bias_attr=bias_attr, + param_attr=param_attr, + act=LinearActivation()) + + hidden_bn = batch_norm_layer( + hidden2, + act=ReluActivation(), + name="dis_hidden_bn", + bias_attr=bias_attr, + param_attr=ParamAttr( + is_static=is_generator_training, initial_mean=1.0, + initial_std=0.02), + use_global_stats=False) + + return fc_layer( + input=hidden_bn, + name="dis_prob", + size=2, + bias_attr=bias_attr, + param_attr=param_attr, + act=SoftmaxActivation()) + def generator(noise): """ generator generates a sample given noise """ param_attr = ParamAttr(is_static=is_discriminator_training) - bias_attr = ParamAttr(is_static=is_discriminator_training, - initial_mean=1.0, - initial_std=0) - - hidden = fc_layer(input=noise, - name="gen_layer_hidden", - size=hidden_dim, - bias_attr=bias_attr, - param_attr=param_attr, - act=ReluActivation()) - - hidden2 = fc_layer(input=hidden, name="gen_hidden2", size=hidden_dim, - bias_attr=bias_attr, - param_attr=param_attr, - act=LinearActivation()) - - hidden_bn = batch_norm_layer(hidden2, - act=ReluActivation(), - name="gen_layer_hidden_bn", - bias_attr=bias_attr, - param_attr=ParamAttr(is_static=is_discriminator_training, - initial_mean=1.0, - initial_std=0.02), - use_global_stats=False) - - return fc_layer(input=hidden_bn, - name="gen_layer1", - size=sample_dim, - bias_attr=bias_attr, - param_attr=param_attr, - act=LinearActivation()) + bias_attr = ParamAttr( + is_static=is_discriminator_training, initial_mean=1.0, initial_std=0) + + hidden = fc_layer( + input=noise, + name="gen_layer_hidden", + size=hidden_dim, + bias_attr=bias_attr, + param_attr=param_attr, + act=ReluActivation()) + + hidden2 = fc_layer( + input=hidden, + name="gen_hidden2", + size=hidden_dim, + bias_attr=bias_attr, + param_attr=param_attr, + act=LinearActivation()) + + hidden_bn = batch_norm_layer( + hidden2, + act=ReluActivation(), + name="gen_layer_hidden_bn", + bias_attr=bias_attr, + param_attr=ParamAttr( + is_static=is_discriminator_training, + initial_mean=1.0, + initial_std=0.02), + use_global_stats=False) + + return fc_layer( + input=hidden_bn, + name="gen_layer1", + size=sample_dim, + bias_attr=bias_attr, + param_attr=param_attr, + act=LinearActivation()) + if is_generator_training: noise = data_layer(name="noise", size=noise_dim) @@ -126,7 +142,8 @@ if is_generator_training or is_discriminator_training: label = data_layer(name="label", size=1) prob = discriminator(sample) cost = cross_entropy(input=prob, label=label) - classification_error_evaluator(input=prob, label=label, name=mode+'_error') + classification_error_evaluator( + input=prob, label=label, name=mode + '_error') outputs(cost) if is_generator: diff --git a/demo/gan/gan_conf_image.py b/demo/gan/gan_conf_image.py index 5c2b140537..f89a4e706c 100644 --- a/demo/gan/gan_conf_image.py +++ b/demo/gan/gan_conf_image.py @@ -15,10 +15,9 @@ from paddle.trainer_config_helpers import * mode = get_config_arg("mode", str, "generator") dataSource = get_config_arg("data", str, "mnist") -assert mode in set(["generator", - "discriminator", - "generator_training", - "discriminator_training"]) +assert mode in set([ + "generator", "discriminator", "generator_training", "discriminator_training" +]) is_generator_training = mode == "generator_training" is_discriminator_training = mode == "discriminator_training" @@ -36,24 +35,33 @@ noise_dim = 100 gf_dim = 64 df_dim = 64 if dataSource == "mnist": - sample_dim = 28 # image dim - c_dim = 1 # image color + sample_dim = 28 # image dim + c_dim = 1 # image color else: sample_dim = 32 c_dim = 3 -s2, s4 = int(sample_dim/2), int(sample_dim/4), -s8, s16 = int(sample_dim/8), int(sample_dim/16) +s2, s4 = int(sample_dim / 2), int(sample_dim / 4), +s8, s16 = int(sample_dim / 8), int(sample_dim / 16) settings( batch_size=128, learning_rate=2e-4, - learning_method=AdamOptimizer(beta1=0.5) -) + learning_method=AdamOptimizer(beta1=0.5)) -def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name, - param_attr, bias_attr, param_attr_bn, bn, trans=False, - act=ReluActivation()): - + +def conv_bn(input, + channels, + imgSize, + num_filters, + output_x, + stride, + name, + param_attr, + bias_attr, + param_attr_bn, + bn, + trans=False, + act=ReluActivation()): """ conv_bn is a utility function that constructs a convolution/deconv layer with an optional batch_norm layer @@ -63,10 +71,10 @@ def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name, :param trans: whether to use conv (False) or deconv (True) :type trans: bool """ - + # calculate the filter_size and padding size based on the given # imgSize and ouput size - tmp = imgSize - (output_x - 1) * stride + tmp = imgSize - (output_x - 1) * stride if tmp <= 1 or tmp > 5: raise ValueError("conv input-output dimension does not fit") elif tmp <= 3: @@ -76,111 +84,134 @@ def conv_bn(input, channels, imgSize, num_filters, output_x, stride, name, filter_size = tmp padding = 0 - print (imgSize, output_x, stride, filter_size, padding) - + print(imgSize, output_x, stride, filter_size, padding) + if trans: nameApx = "_conv" else: nameApx = "_convt" - + if bn: - conv = img_conv_layer(input, filter_size=filter_size, - num_filters=num_filters, - name=name + nameApx, num_channels=channels, - act=LinearActivation(), groups=1, stride=stride, - padding=padding, bias_attr=bias_attr, - param_attr=param_attr, shared_biases=True, layer_attr=None, - filter_size_y=None, stride_y=None, padding_y=None, - trans=trans) - - conv_bn = batch_norm_layer(conv, - act=act, - name=name + nameApx + "_bn", - bias_attr=bias_attr, - param_attr=param_attr_bn, - use_global_stats=False) - + conv = img_conv_layer( + input, + filter_size=filter_size, + num_filters=num_filters, + name=name + nameApx, + num_channels=channels, + act=LinearActivation(), + groups=1, + stride=stride, + padding=padding, + bias_attr=bias_attr, + param_attr=param_attr, + shared_biases=True, + layer_attr=None, + filter_size_y=None, + stride_y=None, + padding_y=None, + trans=trans) + + conv_bn = batch_norm_layer( + conv, + act=act, + name=name + nameApx + "_bn", + bias_attr=bias_attr, + param_attr=param_attr_bn, + use_global_stats=False) + return conv_bn else: - conv = img_conv_layer(input, filter_size=filter_size, - num_filters=num_filters, - name=name + nameApx, num_channels=channels, - act=act, groups=1, stride=stride, - padding=padding, bias_attr=bias_attr, - param_attr=param_attr, shared_biases=True, layer_attr=None, - filter_size_y=None, stride_y=None, padding_y=None, - trans=trans) + conv = img_conv_layer( + input, + filter_size=filter_size, + num_filters=num_filters, + name=name + nameApx, + num_channels=channels, + act=act, + groups=1, + stride=stride, + padding=padding, + bias_attr=bias_attr, + param_attr=param_attr, + shared_biases=True, + layer_attr=None, + filter_size_y=None, + stride_y=None, + padding_y=None, + trans=trans) return conv - + + def generator(noise): """ generator generates a sample given noise """ - param_attr = ParamAttr(is_static=is_discriminator_training, - initial_mean=0.0, - initial_std=0.02) - bias_attr = ParamAttr(is_static=is_discriminator_training, - initial_mean=0.0, - initial_std=0.0) - - param_attr_bn=ParamAttr(is_static=is_discriminator_training, - initial_mean=1.0, - initial_std=0.02) - - h1 = fc_layer(input=noise, - name="gen_layer_h1", - size=s8 * s8 * gf_dim * 4, - bias_attr=bias_attr, - param_attr=param_attr, - act=LinearActivation()) - - h1_bn = batch_norm_layer(h1, - act=ReluActivation(), - name="gen_layer_h1_bn", - bias_attr=bias_attr, - param_attr=param_attr_bn, - use_global_stats=False) - - h2_bn = conv_bn(h1_bn, - channels=gf_dim*4, - output_x=s8, - num_filters=gf_dim*2, - imgSize=s4, - stride=2, - name="gen_layer_h2", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=True, - trans=True) - - h3_bn = conv_bn(h2_bn, - channels=gf_dim*2, - output_x=s4, - num_filters=gf_dim, - imgSize=s2, - stride=2, - name="gen_layer_h3", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=True, - trans=True) - - - return conv_bn(h3_bn, - channels=gf_dim, - output_x=s2, - num_filters=c_dim, - imgSize=sample_dim, - stride=2, - name="gen_layer_h4", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=False, - trans=True, - act=TanhActivation()) + param_attr = ParamAttr( + is_static=is_discriminator_training, initial_mean=0.0, initial_std=0.02) + bias_attr = ParamAttr( + is_static=is_discriminator_training, initial_mean=0.0, initial_std=0.0) + + param_attr_bn = ParamAttr( + is_static=is_discriminator_training, initial_mean=1.0, initial_std=0.02) + + h1 = fc_layer( + input=noise, + name="gen_layer_h1", + size=s8 * s8 * gf_dim * 4, + bias_attr=bias_attr, + param_attr=param_attr, + act=LinearActivation()) + + h1_bn = batch_norm_layer( + h1, + act=ReluActivation(), + name="gen_layer_h1_bn", + bias_attr=bias_attr, + param_attr=param_attr_bn, + use_global_stats=False) + + h2_bn = conv_bn( + h1_bn, + channels=gf_dim * 4, + output_x=s8, + num_filters=gf_dim * 2, + imgSize=s4, + stride=2, + name="gen_layer_h2", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=True, + trans=True) + + h3_bn = conv_bn( + h2_bn, + channels=gf_dim * 2, + output_x=s4, + num_filters=gf_dim, + imgSize=s2, + stride=2, + name="gen_layer_h3", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=True, + trans=True) + + return conv_bn( + h3_bn, + channels=gf_dim, + output_x=s2, + num_filters=c_dim, + imgSize=sample_dim, + stride=2, + name="gen_layer_h4", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=False, + trans=True, + act=TanhActivation()) def discriminator(sample): @@ -191,58 +222,60 @@ def discriminator(sample): of the sample is from generator and dimension 1 is the probabblity of the sample is from real data. """ - param_attr = ParamAttr(is_static=is_generator_training, - initial_mean=0.0, - initial_std=0.02) - bias_attr = ParamAttr(is_static=is_generator_training, - initial_mean=0.0, - initial_std=0.0) - - param_attr_bn=ParamAttr(is_static=is_generator_training, - initial_mean=1.0, - initial_std=0.02) - - h0 = conv_bn(sample, - channels=c_dim, - imgSize=sample_dim, - num_filters=df_dim, - output_x=s2, - stride=2, - name="dis_h0", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=False) - - h1_bn = conv_bn(h0, - channels=df_dim, - imgSize=s2, - num_filters=df_dim*2, - output_x=s4, - stride=2, - name="dis_h1", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=True) - - h2_bn = conv_bn(h1_bn, - channels=df_dim*2, - imgSize=s4, - num_filters=df_dim*4, - output_x=s8, - stride=2, - name="dis_h2", - param_attr=param_attr, - bias_attr=bias_attr, - param_attr_bn=param_attr_bn, - bn=True) - - return fc_layer(input=h2_bn, name="dis_prob", size=2, - bias_attr=bias_attr, - param_attr=param_attr, - act=SoftmaxActivation()) + param_attr = ParamAttr( + is_static=is_generator_training, initial_mean=0.0, initial_std=0.02) + bias_attr = ParamAttr( + is_static=is_generator_training, initial_mean=0.0, initial_std=0.0) + + param_attr_bn = ParamAttr( + is_static=is_generator_training, initial_mean=1.0, initial_std=0.02) + + h0 = conv_bn( + sample, + channels=c_dim, + imgSize=sample_dim, + num_filters=df_dim, + output_x=s2, + stride=2, + name="dis_h0", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=False) + + h1_bn = conv_bn( + h0, + channels=df_dim, + imgSize=s2, + num_filters=df_dim * 2, + output_x=s4, + stride=2, + name="dis_h1", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=True) + + h2_bn = conv_bn( + h1_bn, + channels=df_dim * 2, + imgSize=s4, + num_filters=df_dim * 4, + output_x=s8, + stride=2, + name="dis_h2", + param_attr=param_attr, + bias_attr=bias_attr, + param_attr_bn=param_attr_bn, + bn=True) + return fc_layer( + input=h2_bn, + name="dis_prob", + size=2, + bias_attr=bias_attr, + param_attr=param_attr, + act=SoftmaxActivation()) if is_generator_training: @@ -250,13 +283,14 @@ if is_generator_training: sample = generator(noise) if is_discriminator_training: - sample = data_layer(name="sample", size=sample_dim * sample_dim*c_dim) + sample = data_layer(name="sample", size=sample_dim * sample_dim * c_dim) if is_generator_training or is_discriminator_training: label = data_layer(name="label", size=1) prob = discriminator(sample) cost = cross_entropy(input=prob, label=label) - classification_error_evaluator(input=prob, label=label, name=mode+'_error') + classification_error_evaluator( + input=prob, label=label, name=mode + '_error') outputs(cost) if is_generator: diff --git a/demo/gan/gan_trainer.py b/demo/gan/gan_trainer.py index a8c1bd0414..4a26c230f7 100644 --- a/demo/gan/gan_trainer.py +++ b/demo/gan/gan_trainer.py @@ -16,7 +16,7 @@ import argparse import random import numpy import cPickle -import sys,os +import sys, os from PIL import Image from paddle.trainer.config_parser import parse_config @@ -24,6 +24,7 @@ from paddle.trainer.config_parser import logger import py_paddle.swig_paddle as api import matplotlib.pyplot as plt + def plot2DScatter(data, outputfile): ''' Plot the data as a 2D scatter plot and save to outputfile @@ -41,9 +42,11 @@ def plot2DScatter(data, outputfile): plt.scatter(x, y) plt.savefig(outputfile, bbox_inches='tight') + def CHECK_EQ(a, b): assert a == b, "a=%s, b=%s" % (a, b) + def copy_shared_parameters(src, dst): ''' copy the parameters from src to dst @@ -52,11 +55,9 @@ def copy_shared_parameters(src, dst): :param dst: the destination of the parameters :type dst: GradientMachine ''' - src_params = [src.getParameter(i) - for i in xrange(src.getParameterSize())] + src_params = [src.getParameter(i) for i in xrange(src.getParameterSize())] src_params = dict([(p.getName(), p) for p in src_params]) - for i in xrange(dst.getParameterSize()): dst_param = dst.getParameter(i) src_param = src_params.get(dst_param.getName(), None) @@ -67,15 +68,17 @@ def copy_shared_parameters(src, dst): CHECK_EQ(len(src_value), len(dst_value)) dst_value.copyFrom(src_value) dst_param.setValueUpdated() - + + def print_parameters(src): - src_params = [src.getParameter(i) - for i in xrange(src.getParameterSize())] + src_params = [src.getParameter(i) for i in xrange(src.getParameterSize())] print "***************" for p in src_params: print "Name is %s" % p.getName() - print "value is %s \n" % p.getBuf(api.PARAMETER_VALUE).copyToNumpyArray() + print "value is %s \n" % p.getBuf(api.PARAMETER_VALUE).copyToNumpyArray( + ) + def load_mnist_data(imageFile): f = open(imageFile, "rb") @@ -86,33 +89,36 @@ def load_mnist_data(imageFile): n = 60000 else: n = 10000 - - data = numpy.fromfile(f, 'ubyte', count=n*28*28).reshape((n, 28*28)) + + data = numpy.fromfile(f, 'ubyte', count=n * 28 * 28).reshape((n, 28 * 28)) data = data / 255.0 * 2.0 - 1.0 f.close() return data.astype('float32') + def load_cifar_data(cifar_path): batch_size = 10000 - data = numpy.zeros((5*batch_size, 32*32*3), dtype = "float32") + data = numpy.zeros((5 * batch_size, 32 * 32 * 3), dtype="float32") for i in range(1, 6): file = cifar_path + "/data_batch_" + str(i) fo = open(file, 'rb') dict = cPickle.load(fo) fo.close() - data[(i - 1)*batch_size:(i*batch_size), :] = dict["data"] - + data[(i - 1) * batch_size:(i * batch_size), :] = dict["data"] + data = data / 255.0 * 2.0 - 1.0 return data + # synthesize 2-D uniform data def load_uniform_data(): data = numpy.random.rand(1000000, 2).astype('float32') return data + def merge(images, size): - if images.shape[1] == 28*28: + if images.shape[1] == 28 * 28: h, w, c = 28, 28, 1 else: h, w, c = 32, 32, 3 @@ -124,6 +130,7 @@ def merge(images, size): ((images[idx, :].reshape((h, w, c), order="F").transpose(1, 0, 2) + 1.0) / 2.0 * 255.0) return img.astype('uint8') + def save_images(images, path): merged_img = merge(images, [8, 8]) if merged_img.shape[2] == 1: @@ -131,14 +138,17 @@ def save_images(images, path): else: im = Image.fromarray(merged_img, mode="RGB") im.save(path) - + + def get_real_samples(batch_size, data_np): - return data_np[numpy.random.choice(data_np.shape[0], batch_size, - replace=False),:] - + return data_np[numpy.random.choice( + data_np.shape[0], batch_size, replace=False), :] + + def get_noise(batch_size, noise_dim): return numpy.random.normal(size=(batch_size, noise_dim)).astype('float32') + def get_fake_samples(generator_machine, batch_size, noise): gen_inputs = api.Arguments.createArguments(1) gen_inputs.setSlotValue(0, api.Matrix.createDenseFromNumpy(noise)) @@ -147,12 +157,14 @@ def get_fake_samples(generator_machine, batch_size, noise): fake_samples = gen_outputs.getSlotValue(0).copyToNumpyMat() return fake_samples + def get_training_loss(training_machine, inputs): outputs = api.Arguments.createArguments(0) training_machine.forward(inputs, outputs, api.PASS_TEST) loss = outputs.getSlotValue(0).copyToNumpyMat() return numpy.mean(loss) + def prepare_discriminator_data_batch_pos(batch_size, data_np): real_samples = get_real_samples(batch_size, data_np) labels = numpy.ones(batch_size, dtype='int32') @@ -161,6 +173,7 @@ def prepare_discriminator_data_batch_pos(batch_size, data_np): inputs.setSlotIds(1, api.IVector.createVectorFromNumpy(labels)) return inputs + def prepare_discriminator_data_batch_neg(generator_machine, batch_size, noise): fake_samples = get_fake_samples(generator_machine, batch_size, noise) labels = numpy.zeros(batch_size, dtype='int32') @@ -169,6 +182,7 @@ def prepare_discriminator_data_batch_neg(generator_machine, batch_size, noise): inputs.setSlotIds(1, api.IVector.createVectorFromNumpy(labels)) return inputs + def prepare_generator_data_batch(batch_size, noise): label = numpy.ones(batch_size, dtype='int32') inputs = api.Arguments.createArguments(2) @@ -193,10 +207,9 @@ def get_layer_size(model_conf, layer_name): def main(): parser = argparse.ArgumentParser() parser.add_argument("-d", "--data_source", help="mnist or cifar or uniform") - parser.add_argument("--use_gpu", default="1", - help="1 means use gpu for training") - parser.add_argument("--gpu_id", default="0", - help="the gpu_id parameter") + parser.add_argument( + "--use_gpu", default="1", help="1 means use gpu for training") + parser.add_argument("--gpu_id", default="0", help="the gpu_id parameter") args = parser.parse_args() data_source = args.data_source use_gpu = args.use_gpu @@ -208,30 +221,32 @@ def main(): if not os.path.exists("./%s_params/" % data_source): os.makedirs("./%s_params/" % data_source) - - api.initPaddle('--use_gpu=' + use_gpu, '--dot_period=10', '--log_period=100', - '--gpu_id=' + args.gpu_id, '--save_dir=' + "./%s_params/" % data_source) - + + api.initPaddle('--use_gpu=' + use_gpu, '--dot_period=10', + '--log_period=100', '--gpu_id=' + args.gpu_id, + '--save_dir=' + "./%s_params/" % data_source) + if data_source == "uniform": conf = "gan_conf.py" num_iter = 10000 else: conf = "gan_conf_image.py" num_iter = 1000 - + gen_conf = parse_config(conf, "mode=generator_training,data=" + data_source) - dis_conf = parse_config(conf, "mode=discriminator_training,data=" + data_source) + dis_conf = parse_config(conf, + "mode=discriminator_training,data=" + data_source) generator_conf = parse_config(conf, "mode=generator,data=" + data_source) batch_size = dis_conf.opt_config.batch_size noise_dim = get_layer_size(gen_conf.model_config, "noise") - + if data_source == "mnist": data_np = load_mnist_data("./data/mnist_data/train-images-idx3-ubyte") elif data_source == "cifar": data_np = load_cifar_data("./data/cifar-10-batches-py/") else: data_np = load_uniform_data() - + # this creates a gradient machine for discriminator dis_training_machine = api.GradientMachine.createFromConfigProto( dis_conf.model_config) @@ -244,26 +259,24 @@ def main(): logger.info(str(generator_conf.model_config)) generator_machine = api.GradientMachine.createFromConfigProto( generator_conf.model_config) - - dis_trainer = api.Trainer.create( - dis_conf, dis_training_machine) - gen_trainer = api.Trainer.create( - gen_conf, gen_training_machine) - + dis_trainer = api.Trainer.create(dis_conf, dis_training_machine) + + gen_trainer = api.Trainer.create(gen_conf, gen_training_machine) + dis_trainer.startTrain() gen_trainer.startTrain() - + # Sync parameters between networks (GradientMachine) at the beginning copy_shared_parameters(gen_training_machine, dis_training_machine) copy_shared_parameters(gen_training_machine, generator_machine) - + # constrain that either discriminator or generator can not be trained # consecutively more than MAX_strike times curr_train = "dis" curr_strike = 0 MAX_strike = 5 - + for train_pass in xrange(100): dis_trainer.startTrainPass() gen_trainer.startTrainPass() @@ -272,23 +285,25 @@ def main(): noise = get_noise(batch_size, noise_dim) data_batch_dis_pos = prepare_discriminator_data_batch_pos( batch_size, data_np) - dis_loss_pos = get_training_loss(dis_training_machine, data_batch_dis_pos) - + dis_loss_pos = get_training_loss(dis_training_machine, + data_batch_dis_pos) + data_batch_dis_neg = prepare_discriminator_data_batch_neg( generator_machine, batch_size, noise) - dis_loss_neg = get_training_loss(dis_training_machine, data_batch_dis_neg) - + dis_loss_neg = get_training_loss(dis_training_machine, + data_batch_dis_neg) + dis_loss = (dis_loss_pos + dis_loss_neg) / 2.0 - + # Do forward pass in generator to get the gen_loss - data_batch_gen = prepare_generator_data_batch( - batch_size, noise) + data_batch_gen = prepare_generator_data_batch(batch_size, noise) gen_loss = get_training_loss(gen_training_machine, data_batch_gen) - + if i % 100 == 0: - print "d_pos_loss is %s d_neg_loss is %s" % (dis_loss_pos, dis_loss_neg) + print "d_pos_loss is %s d_neg_loss is %s" % (dis_loss_pos, + dis_loss_neg) print "d_loss is %s g_loss is %s" % (dis_loss, gen_loss) - + # Decide which network to train based on the training history # And the relative size of the loss if (not (curr_train == "dis" and curr_strike == MAX_strike)) and \ @@ -297,11 +312,12 @@ def main(): curr_strike += 1 else: curr_train = "dis" - curr_strike = 1 + curr_strike = 1 dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_neg) - dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_pos) - copy_shared_parameters(dis_training_machine, gen_training_machine) - + dis_trainer.trainOneDataBatch(batch_size, data_batch_dis_pos) + copy_shared_parameters(dis_training_machine, + gen_training_machine) + else: if curr_train == "gen": curr_strike += 1 @@ -311,19 +327,23 @@ def main(): gen_trainer.trainOneDataBatch(batch_size, data_batch_gen) # TODO: add API for paddle to allow true parameter sharing between different GradientMachines # so that we do not need to copy shared parameters. - copy_shared_parameters(gen_training_machine, dis_training_machine) + copy_shared_parameters(gen_training_machine, + dis_training_machine) copy_shared_parameters(gen_training_machine, generator_machine) - + dis_trainer.finishTrainPass() gen_trainer.finishTrainPass() # At the end of each pass, save the generated samples/images fake_samples = get_fake_samples(generator_machine, batch_size, noise) if data_source == "uniform": - plot2DScatter(fake_samples, "./%s_samples/train_pass%s.png" % (data_source, train_pass)) + plot2DScatter(fake_samples, "./%s_samples/train_pass%s.png" % + (data_source, train_pass)) else: - save_images(fake_samples, "./%s_samples/train_pass%s.png" % (data_source, train_pass)) + save_images(fake_samples, "./%s_samples/train_pass%s.png" % + (data_source, train_pass)) dis_trainer.finishTrain() gen_trainer.finishTrain() + if __name__ == '__main__': main() diff --git a/demo/quick_start/trainer_config.resnet-lstm.py b/demo/quick_start/trainer_config.resnet-lstm.py index 5bed925d84..89a837abb7 100644 --- a/demo/quick_start/trainer_config.resnet-lstm.py +++ b/demo/quick_start/trainer_config.resnet-lstm.py @@ -13,7 +13,6 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - """ This configuration is a demonstration of how to implement the stacked LSTM with residual connections, i.e. an LSTM layer takes the sum of the hidden states @@ -46,11 +45,12 @@ is_predict = get_config_arg('is_predict', bool, False) trn = 'data/train.list' if not is_predict else None tst = 'data/test.list' if not is_predict else 'data/pred.list' process = 'process' if not is_predict else 'process_predict' -define_py_data_sources2(train_list=trn, - test_list=tst, - module="dataprovider_emb", - obj=process, - args={"dictionary": word_dict}) +define_py_data_sources2( + train_list=trn, + test_list=tst, + module="dataprovider_emb", + obj=process, + args={"dictionary": word_dict}) batch_size = 128 if not is_predict else 1 settings( @@ -58,10 +58,9 @@ settings( learning_rate=2e-3, learning_method=AdamOptimizer(), regularization=L2Regularization(8e-4), - gradient_clipping_threshold=25 -) + gradient_clipping_threshold=25) -bias_attr = ParamAttr(initial_std=0.,l2_rate=0.) +bias_attr = ParamAttr(initial_std=0., l2_rate=0.) data = data_layer(name="word", size=len(word_dict)) emb = embedding_layer(input=data, size=128) @@ -73,17 +72,15 @@ for i in range(3): # The input to the current layer is the sum of the hidden state # and input of the previous layer. current_input = addto_layer(input=[previous_input, previous_hidden_state]) - hidden_state = simple_lstm(input=current_input, size=128, - lstm_cell_attr=ExtraAttr(drop_rate=0.1)) + hidden_state = simple_lstm( + input=current_input, size=128, lstm_cell_attr=ExtraAttr(drop_rate=0.1)) previous_input, previous_hidden_state = current_input, hidden_state lstm = previous_hidden_state lstm_last = pooling_layer(input=lstm, pooling_type=MaxPooling()) -output = fc_layer(input=lstm_last, size=2, - bias_attr=bias_attr, - act=SoftmaxActivation()) - +output = fc_layer( + input=lstm_last, size=2, bias_attr=bias_attr, act=SoftmaxActivation()) if is_predict: maxid = maxid_layer(output) diff --git a/demo/semantic_role_labeling/data/extract_dict_feature.py b/demo/semantic_role_labeling/data/extract_dict_feature.py index 123df022f5..a02a49a86e 100644 --- a/demo/semantic_role_labeling/data/extract_dict_feature.py +++ b/demo/semantic_role_labeling/data/extract_dict_feature.py @@ -33,7 +33,7 @@ def extract_dict_features(pair_file, feature_file): ctx_n1 = sentence_list[verb_index - 1] else: ctx_n1 = 'bos' - + if verb_index > 1: mark[verb_index - 2] = 1 ctx_n2 = sentence_list[verb_index - 2] @@ -48,7 +48,7 @@ def extract_dict_features(pair_file, feature_file): ctx_p1 = sentence_list[verb_index + 1] else: ctx_p1 = 'eos' - + if verb_index < len(labels_list) - 3: mark[verb_index + 2] = 1 ctx_p2 = sentence_list[verb_index + 2] @@ -69,7 +69,6 @@ def extract_dict_features(pair_file, feature_file): feature_out.write(feature_str + '\n') - if __name__ == '__main__': usage = '-p pair_file -f feature_file' diff --git a/demo/semantic_role_labeling/data/extract_pairs.py b/demo/semantic_role_labeling/data/extract_pairs.py index 2d0d535c53..94a8488c16 100644 --- a/demo/semantic_role_labeling/data/extract_pairs.py +++ b/demo/semantic_role_labeling/data/extract_pairs.py @@ -66,8 +66,8 @@ def transform_labels(sentences, labels): else: verb_list = [] for x in labels[i][0]: - if x !='-': - verb_list.append(x) + if x != '-': + verb_list.append(x) for j in xrange(1, len(labels[i])): label_list = labels[i][j] @@ -93,7 +93,7 @@ def transform_labels(sentences, labels): is_in_bracket = True else: print 'error:', ll - sen_lab_pair.append((sentences[i], verb_list[j-1], label_seq)) + sen_lab_pair.append((sentences[i], verb_list[j - 1], label_seq)) return sen_lab_pair @@ -103,7 +103,7 @@ def write_file(sen_lab_pair, output_file): sentence = x[0] label_seq = ' '.join(x[2]) assert len(sentence.split()) == len(x[2]) - fout.write(sentence + '\t' + x[1]+'\t' +label_seq + '\n') + fout.write(sentence + '\t' + x[1] + '\t' + label_seq + '\n') if __name__ == '__main__': diff --git a/demo/semantic_role_labeling/dataprovider.py b/demo/semantic_role_labeling/dataprovider.py index d12f10bfcb..042cd4e7a9 100644 --- a/demo/semantic_role_labeling/dataprovider.py +++ b/demo/semantic_role_labeling/dataprovider.py @@ -21,7 +21,7 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs): settings.word_dict = word_dict settings.label_dict = label_dict settings.predicate_dict = predicate_dict - + #all inputs are integral and sequential type settings.slots = [ integer_value_sequence(len(word_dict)), @@ -29,25 +29,28 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs): integer_value_sequence(len(word_dict)), integer_value_sequence(len(word_dict)), integer_value_sequence(len(word_dict)), - integer_value_sequence(len(word_dict)), - integer_value_sequence(len(predicate_dict)), - integer_value_sequence(2), + integer_value_sequence(len(word_dict)), + integer_value_sequence(len(predicate_dict)), integer_value_sequence(2), integer_value_sequence(len(label_dict)) ] def get_batch_size(yeild_data): return len(yeild_data[0]) - -@provider(init_hook=hook, should_shuffle=True, calc_batch_size=get_batch_size, - can_over_batch_size=False, cache=CacheType.CACHE_PASS_IN_MEM) + +@provider( + init_hook=hook, + should_shuffle=True, + calc_batch_size=get_batch_size, + can_over_batch_size=False, + cache=CacheType.CACHE_PASS_IN_MEM) def process(settings, file_name): with open(file_name, 'r') as fdata: for line in fdata: sentence, predicate, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2, mark, label = \ line.strip().split('\t') - + words = sentence.split() sen_len = len(words) word_slot = [settings.word_dict.get(w, UNK_IDX) for w in words] diff --git a/demo/semantic_role_labeling/db_lstm.py b/demo/semantic_role_labeling/db_lstm.py index 75946bd72e..04e2a559b1 100644 --- a/demo/semantic_role_labeling/db_lstm.py +++ b/demo/semantic_role_labeling/db_lstm.py @@ -20,7 +20,7 @@ from paddle.trainer_config_helpers import * #file paths word_dict_file = './data/wordDict.txt' label_dict_file = './data/targetDict.txt' -predicate_file= './data/verbDict.txt' +predicate_file = './data/verbDict.txt' train_list_file = './data/train.list' test_list_file = './data/test.list' @@ -47,7 +47,6 @@ if not is_predict: w = line.strip() predicate_dict[w] = i - if is_test: train_list_file = None @@ -57,9 +56,11 @@ if not is_predict: test_list=test_list_file, module='dataprovider', obj='process', - args={'word_dict': word_dict, - 'label_dict': label_dict, - 'predicate_dict': predicate_dict }) + args={ + 'word_dict': word_dict, + 'label_dict': label_dict, + 'predicate_dict': predicate_dict + }) word_dict_len = len(word_dict) label_dict_len = len(label_dict) @@ -77,24 +78,16 @@ mark_dim = 5 hidden_dim = 512 depth = 8 - - ########################### Optimizer ####################################### - settings( batch_size=150, learning_method=MomentumOptimizer(momentum=0), learning_rate=2e-2, regularization=L2Regularization(8e-4), is_async=False, - model_average=ModelAverage(average_window=0.5, - max_average_window=10000), - -) - - - + model_average=ModelAverage( + average_window=0.5, max_average_window=10000), ) ####################################### network ############################## #8 features and 1 target @@ -108,22 +101,28 @@ ctx_p1 = data_layer(name='ctx_p1_data', size=word_dict_len) ctx_p2 = data_layer(name='ctx_p2_data', size=word_dict_len) mark = data_layer(name='mark_data', size=mark_dict_len) - if not is_predict: target = data_layer(name='target', size=label_dict_len) - -default_std=1/math.sqrt(hidden_dim)/3.0 +default_std = 1 / math.sqrt(hidden_dim) / 3.0 emb_para = ParameterAttribute(name='emb', initial_std=0., learning_rate=0.) std_0 = ParameterAttribute(initial_std=0.) -std_default = ParameterAttribute(initial_std=default_std) - -predicate_embedding = embedding_layer(size=word_dim, input=predicate, param_attr=ParameterAttribute(name='vemb',initial_std=default_std)) -mark_embedding = embedding_layer(name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0) - -word_input=[word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2] -emb_layers = [embedding_layer(size=word_dim, input=x, param_attr=emb_para) for x in word_input] +std_default = ParameterAttribute(initial_std=default_std) + +predicate_embedding = embedding_layer( + size=word_dim, + input=predicate, + param_attr=ParameterAttribute( + name='vemb', initial_std=default_std)) +mark_embedding = embedding_layer( + name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0) + +word_input = [word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2] +emb_layers = [ + embedding_layer( + size=word_dim, input=x, param_attr=emb_para) for x in word_input +] emb_layers.append(predicate_embedding) emb_layers.append(mark_embedding) @@ -131,84 +130,89 @@ hidden_0 = mixed_layer( name='hidden0', size=hidden_dim, bias_attr=std_default, - input=[ full_matrix_projection(input=emb, param_attr=std_default ) for emb in emb_layers ]) - + input=[ + full_matrix_projection( + input=emb, param_attr=std_default) for emb in emb_layers + ]) mix_hidden_lr = 1e-3 lstm_para_attr = ParameterAttribute(initial_std=0.0, learning_rate=1.0) -hidden_para_attr = ParameterAttribute(initial_std=default_std, learning_rate=mix_hidden_lr) - -lstm_0 = lstmemory(name='lstm0', - input=hidden_0, - act=ReluActivation(), - gate_act=SigmoidActivation(), - state_act=SigmoidActivation(), - bias_attr=std_0, - param_attr=lstm_para_attr) +hidden_para_attr = ParameterAttribute( + initial_std=default_std, learning_rate=mix_hidden_lr) + +lstm_0 = lstmemory( + name='lstm0', + input=hidden_0, + act=ReluActivation(), + gate_act=SigmoidActivation(), + state_act=SigmoidActivation(), + bias_attr=std_0, + param_attr=lstm_para_attr) #stack L-LSTM and R-LSTM with direct edges input_tmp = [hidden_0, lstm_0] - for i in range(1, depth): - mix_hidden = mixed_layer(name='hidden'+str(i), - size=hidden_dim, - bias_attr=std_default, - input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr), - full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr) - ] - ) - - lstm = lstmemory(name='lstm'+str(i), - input=mix_hidden, - act=ReluActivation(), - gate_act=SigmoidActivation(), - state_act=SigmoidActivation(), - reverse=((i % 2)==1), - bias_attr=std_0, - param_attr=lstm_para_attr) + mix_hidden = mixed_layer( + name='hidden' + str(i), + size=hidden_dim, + bias_attr=std_default, + input=[ + full_matrix_projection( + input=input_tmp[0], param_attr=hidden_para_attr), + full_matrix_projection( + input=input_tmp[1], param_attr=lstm_para_attr) + ]) + + lstm = lstmemory( + name='lstm' + str(i), + input=mix_hidden, + act=ReluActivation(), + gate_act=SigmoidActivation(), + state_act=SigmoidActivation(), + reverse=((i % 2) == 1), + bias_attr=std_0, + param_attr=lstm_para_attr) input_tmp = [mix_hidden, lstm] -feature_out = mixed_layer(name='output', - size=label_dict_len, - bias_attr=std_default, - input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr), - full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr) - ], - ) - - +feature_out = mixed_layer( + name='output', + size=label_dict_len, + bias_attr=std_default, + input=[ + full_matrix_projection( + input=input_tmp[0], param_attr=hidden_para_attr), + full_matrix_projection( + input=input_tmp[1], param_attr=lstm_para_attr) + ], ) if not is_predict: - crf_l = crf_layer( name = 'crf', - size = label_dict_len, - input = feature_out, - label = target, - param_attr=ParameterAttribute(name='crfw',initial_std=default_std, learning_rate=mix_hidden_lr) - - ) - - - crf_dec_l = crf_decoding_layer(name = 'crf_dec_l', - size = label_dict_len, - input = feature_out, - label = target, - param_attr=ParameterAttribute(name='crfw') - ) - + crf_l = crf_layer( + name='crf', + size=label_dict_len, + input=feature_out, + label=target, + param_attr=ParameterAttribute( + name='crfw', initial_std=default_std, learning_rate=mix_hidden_lr)) + + crf_dec_l = crf_decoding_layer( + name='crf_dec_l', + size=label_dict_len, + input=feature_out, + label=target, + param_attr=ParameterAttribute(name='crfw')) eval = sum_evaluator(input=crf_dec_l) - + outputs(crf_l) else: - crf_dec_l = crf_decoding_layer(name = 'crf_dec_l', - size = label_dict_len, - input = feature_out, - param_attr=ParameterAttribute(name='crfw') - ) + crf_dec_l = crf_decoding_layer( + name='crf_dec_l', + size=label_dict_len, + input=feature_out, + param_attr=ParameterAttribute(name='crfw')) outputs(crf_dec_l) - diff --git a/demo/semantic_role_labeling/predict.py b/demo/semantic_role_labeling/predict.py index 15145fafce..372fd090b6 100644 --- a/demo/semantic_role_labeling/predict.py +++ b/demo/semantic_role_labeling/predict.py @@ -26,7 +26,8 @@ UNK_IDX = 0 class Prediction(): - def __init__(self, train_conf, dict_file, model_dir, label_file, predicate_dict_file): + def __init__(self, train_conf, dict_file, model_dir, label_file, + predicate_dict_file): """ train_conf: trainer configure. dict_file: word dictionary file name. @@ -35,7 +36,7 @@ class Prediction(): self.dict = {} self.labels = {} - self.predicate_dict={} + self.predicate_dict = {} self.labels_reverse = {} self.load_dict_label(dict_file, label_file, predicate_dict_file) @@ -44,25 +45,18 @@ class Prediction(): len_pred = len(self.predicate_dict) conf = parse_config( - train_conf, - 'dict_len=' + str(len_dict) + - ',label_len=' + str(len_label) + - ',pred_len=' + str(len_pred) + - ',is_predict=True') + train_conf, 'dict_len=' + str(len_dict) + ',label_len=' + + str(len_label) + ',pred_len=' + str(len_pred) + ',is_predict=True') self.network = swig_paddle.GradientMachine.createFromConfigProto( conf.model_config) self.network.loadParameters(model_dir) slots = [ - integer_value_sequence(len_dict), - integer_value_sequence(len_dict), - integer_value_sequence(len_dict), - integer_value_sequence(len_dict), - integer_value_sequence(len_dict), - integer_value_sequence(len_dict), - integer_value_sequence(len_pred), - integer_value_sequence(2) - ] + integer_value_sequence(len_dict), integer_value_sequence(len_dict), + integer_value_sequence(len_dict), integer_value_sequence(len_dict), + integer_value_sequence(len_dict), integer_value_sequence(len_dict), + integer_value_sequence(len_pred), integer_value_sequence(2) + ] self.converter = DataProviderConverter(slots) def load_dict_label(self, dict_file, label_file, predicate_dict_file): @@ -78,6 +72,7 @@ class Prediction(): for line_count, line in enumerate(open(predicate_dict_file, 'r')): self.predicate_dict[line.strip()] = line_count + def get_data(self, data_file): """ Get input data of paddle format. @@ -88,9 +83,10 @@ class Prediction(): ).split('\t') words = sentence.split() sen_len = len(words) - + word_slot = [self.dict.get(w, UNK_IDX) for w in words] - predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX)] * sen_len + predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX) + ] * sen_len ctx_n2_slot = [self.dict.get(ctx_n2, UNK_IDX)] * sen_len ctx_n1_slot = [self.dict.get(ctx_n1, UNK_IDX)] * sen_len ctx_0_slot = [self.dict.get(ctx_0, UNK_IDX)] * sen_len @@ -99,7 +95,7 @@ class Prediction(): marks = mark.split() mark_slot = [int(w) for w in marks] - + yield word_slot, ctx_n2_slot, ctx_n1_slot, \ ctx_0_slot, ctx_p1_slot, ctx_p2_slot, predicate_slot, mark_slot @@ -123,8 +119,9 @@ class Prediction(): def option_parser(): - usage = ("python predict.py -c config -w model_dir " - "-d word dictionary -l label_file -i input_file -p pred_dict_file") + usage = ( + "python predict.py -c config -w model_dir " + "-d word dictionary -l label_file -i input_file -p pred_dict_file") parser = OptionParser(usage="usage: %s [options]" % usage) parser.add_option( "-c", @@ -187,8 +184,9 @@ def main(): output_file = options.output_file swig_paddle.initPaddle("--use_gpu=0") - predict = Prediction(train_conf, dict_file, model_path, label_file, predict_dict_file) - predict.predict(data_file,output_file) + predict = Prediction(train_conf, dict_file, model_path, label_file, + predict_dict_file) + predict.predict(data_file, output_file) if __name__ == '__main__': diff --git a/doc_cn/cluster/k8s/distributed_training_on_kubernetes.md b/doc_cn/cluster/k8s/distributed_training_on_kubernetes.md index d9ed431ec0..64f8fd4b43 100644 --- a/doc_cn/cluster/k8s/distributed_training_on_kubernetes.md +++ b/doc_cn/cluster/k8s/distributed_training_on_kubernetes.md @@ -306,4 +306,4 @@ I1116 09:10:18.019069 50 ParameterClient2.cpp:122] pserver 2 192.168.223.143: I1116 09:10:18.019492 50 ParameterClient2.cpp:122] pserver 3 192.168.223.143:7165 I1116 09:10:18.019716 50 ParameterClient2.cpp:122] pserver 4 192.168.129.71:7164 I1116 09:10:18.019836 50 ParameterClient2.cpp:122] pserver 5 192.168.129.71:7165 -``` \ No newline at end of file +``` diff --git a/doc_cn/cluster/k8s/job.yaml b/doc_cn/cluster/k8s/job.yaml index 1e0ac464b2..488aad0bed 100644 --- a/doc_cn/cluster/k8s/job.yaml +++ b/doc_cn/cluster/k8s/job.yaml @@ -40,4 +40,4 @@ spec: - name: jobpath mountPath: /home/jobpath restartPolicy: Never - \ No newline at end of file + diff --git a/doc_cn/cluster/k8s/start_paddle.py b/doc_cn/cluster/k8s/start_paddle.py index 6a46161410..df00d82919 100755 --- a/doc_cn/cluster/k8s/start_paddle.py +++ b/doc_cn/cluster/k8s/start_paddle.py @@ -19,7 +19,6 @@ import socket import os import argparse - # configuration for cluster API = "/api/v1/namespaces/" JOBSELECTOR = "labelSelector=job-name=" @@ -145,8 +144,8 @@ def startPaddle(idMap={}, train_args_dict=None): if __name__ == '__main__': - parser = argparse.ArgumentParser(prog="start_paddle.py", - description='simple tool for k8s') + parser = argparse.ArgumentParser( + prog="start_paddle.py", description='simple tool for k8s') args, train_args_list = parser.parse_known_args() train_args = refine_unknown_args(train_args_list) train_args_dict = dict(zip(train_args[:-1:2], train_args[1::2])) diff --git a/doc_cn/demo/sentiment_analysis/index.rst b/doc_cn/demo/sentiment_analysis/index.rst index 82400b2459..9d7972b219 100644 --- a/doc_cn/demo/sentiment_analysis/index.rst +++ b/doc_cn/demo/sentiment_analysis/index.rst @@ -1,8 +1,8 @@ -情感分析教程 -=========================== - -.. toctree:: - :maxdepth: 3 - :glob: - +情感分析教程 +=========================== + +.. toctree:: + :maxdepth: 3 + :glob: + Training Locally \ No newline at end of file diff --git a/doc_theme/static/js/paddle_doc_init.js b/doc_theme/static/js/paddle_doc_init.js index 5c815a8d3a..153ce30745 100644 --- a/doc_theme/static/js/paddle_doc_init.js +++ b/doc_theme/static/js/paddle_doc_init.js @@ -28,4 +28,4 @@ $(document).ready(function(){ $('.doc-menu-vertical').find('li.current').last().addClass('active'); $('.doc-menu-vertical').perfectScrollbar(); -}); \ No newline at end of file +}); diff --git a/paddle/api/GradientMachine.cpp b/paddle/api/GradientMachine.cpp index c1b546dbcb..297eaa19bb 100644 --- a/paddle/api/GradientMachine.cpp +++ b/paddle/api/GradientMachine.cpp @@ -15,8 +15,8 @@ limitations under the License. */ #include "PaddleAPI.h" #include "PaddleAPIPrivate.h" -#include "paddle/gserver/gradientmachines/NeuralNetwork.h" #include "Internal.h" +#include "paddle/gserver/gradientmachines/NeuralNetwork.h" std::vector GradientMachine::defaultParamTypes = { PARAMETER_VALUE, PARAMETER_GRADIENT, PARAMETER_MOMENTUM}; diff --git a/paddle/api/Internal.h b/paddle/api/Internal.h index 4a07880d80..d48dd3a04c 100644 --- a/paddle/api/Internal.h +++ b/paddle/api/Internal.h @@ -16,14 +16,13 @@ limitations under the License. */ #include "PaddleAPI.h" -#include #include +#include template void staticCastVector(std::vector* dest, const std::vector& src) { dest->resize(src.size()); - std::transform(src.begin(), - src.end(), - dest->begin(), - [](T1 t) { return static_cast(t); }); + std::transform(src.begin(), src.end(), dest->begin(), [](T1 t) { + return static_cast(t); + }); } diff --git a/paddle/api/Matrix.cpp b/paddle/api/Matrix.cpp index d4c00e7093..7c375e5cfb 100644 --- a/paddle/api/Matrix.cpp +++ b/paddle/api/Matrix.cpp @@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "PaddleAPI.h" #include "paddle/math/Matrix.h" -#include "paddle/math/SparseMatrix.h" -#include "paddle/math/CpuSparseMatrix.h" -#include #include +#include +#include "PaddleAPI.h" +#include "paddle/math/CpuSparseMatrix.h" +#include "paddle/math/SparseMatrix.h" struct MatrixPrivate { std::shared_ptr mat; diff --git a/paddle/api/PaddleAPI.h b/paddle/api/PaddleAPI.h index f3c80e3b06..84a66719c3 100644 --- a/paddle/api/PaddleAPI.h +++ b/paddle/api/PaddleAPI.h @@ -16,8 +16,8 @@ limitations under the License. */ #include #include -#include #include +#include #include #include "paddle/utils/GlobalConstants.h" #include "paddle/utils/TypeDefs.h" diff --git a/paddle/api/Parameter.cpp b/paddle/api/Parameter.cpp index 742ad0679c..4eed00a84a 100644 --- a/paddle/api/Parameter.cpp +++ b/paddle/api/Parameter.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "PaddleAPI.h" #include "paddle/parameter/Parameter.h" +#include "PaddleAPI.h" struct ParameterPrivate { std::shared_ptr sharedPtr; diff --git a/paddle/api/ParameterOptimizer.cpp b/paddle/api/ParameterOptimizer.cpp index 606dccd5ac..21b851dd5e 100644 --- a/paddle/api/ParameterOptimizer.cpp +++ b/paddle/api/ParameterOptimizer.cpp @@ -12,11 +12,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "PaddleAPI.h" -#include "PaddleAPIPrivate.h" #include "paddle/parameter/ParameterOptimizer.h" -#include "Internal.h" #include +#include "Internal.h" +#include "PaddleAPI.h" +#include "PaddleAPIPrivate.h" struct ParameterOptimizerPrivate { std::unique_ptr optimizer; @@ -36,16 +36,13 @@ struct ParameterTraverseCallbackPrivate { size_t sparseId) { std::vector real_vecs; real_vecs.resize(vecs.size()); - std::transform(vecs.begin(), - vecs.end(), - real_vecs.begin(), - [](Vector* v) { - if (v) { - return *(paddle::VectorPtr*)(v->getSharedPtr()); - } else { - return paddle::VectorPtr(); - } - }); + std::transform(vecs.begin(), vecs.end(), real_vecs.begin(), [](Vector* v) { + if (v) { + return *(paddle::VectorPtr*)(v->getSharedPtr()); + } else { + return paddle::VectorPtr(); + } + }); paddle::ParameterConfig& real_conf = *(paddle::ParameterConfig*)(const_cast(conf) diff --git a/paddle/api/SequenceGenerator.cpp b/paddle/api/SequenceGenerator.cpp index 5c65b34f23..8428edc60d 100644 --- a/paddle/api/SequenceGenerator.cpp +++ b/paddle/api/SequenceGenerator.cpp @@ -12,14 +12,14 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include +#include +#include +#include #include "PaddleAPI.h" #include "paddle/gserver/gradientmachines/GradientMachine.h" #include "paddle/parameter/Argument.h" #include "paddle/utils/Flags.h" -#include -#include -#include -#include // used to represent partial sequence struct Path { diff --git a/paddle/api/Trainer.cpp b/paddle/api/Trainer.cpp index 9aeb874bdc..59b47d4b1c 100644 --- a/paddle/api/Trainer.cpp +++ b/paddle/api/Trainer.cpp @@ -16,12 +16,12 @@ limitations under the License. */ #include "PaddleAPIPrivate.h" #include -#include #include +#include +#include "paddle/gserver/gradientmachines/NeuralNetwork.h" #include "paddle/trainer/ParamUtil.h" #include "paddle/trainer/Trainer.h" -#include "paddle/gserver/gradientmachines/NeuralNetwork.h" #include "paddle/trainer/TrainerInternal.h" #include "paddle/utils/Flags.h" diff --git a/paddle/api/Util.cpp b/paddle/api/Util.cpp index 0c9c048099..c3f739568f 100644 --- a/paddle/api/Util.cpp +++ b/paddle/api/Util.cpp @@ -14,16 +14,16 @@ limitations under the License. */ #include "PaddleAPI.h" -#include "paddle/utils/Util.h" -#include "paddle/utils/PythonUtil.h" -#include "paddle/utils/Flags.h" -#include "paddle/utils/Excepts.h" #include "paddle/parameter/Parameter.h" +#include "paddle/utils/Excepts.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/PythonUtil.h" +#include "paddle/utils/Util.h" #include +#include #include #include -#include void initPaddle(int argc, char** argv) { paddle::initMain(argc, argv); diff --git a/paddle/api/Vector.cpp b/paddle/api/Vector.cpp index 4f3ab7de60..874f2fd044 100644 --- a/paddle/api/Vector.cpp +++ b/paddle/api/Vector.cpp @@ -282,7 +282,7 @@ FloatArray Vector::getData() const { } void Vector::copyFrom(Vector* src) throw(RangeError) { - if (src->m->vec->getSize() != m->vec->getSize()) { + if (src->m->vec->getSize() != m->vec->getSize()) { throw RangeError(); } m->vec->copyFrom(*src->m->vec); diff --git a/paddle/api/test/testMatrix.py b/paddle/api/test/testMatrix.py index f76f84d2e1..37666bdccc 100644 --- a/paddle/api/test/testMatrix.py +++ b/paddle/api/test/testMatrix.py @@ -100,11 +100,12 @@ class TestMatrix(unittest.TestCase): for a, e in zip(gpu_m.getData(), [1.0, 3.23, 3.0, 4.0, 5.0, 6.0]): self.assertAlmostEqual(a, e) - + def test_numpy(self): numpy_mat = np.matrix([[1, 2], [3, 4], [5, 6]], dtype="float32") m = swig_paddle.Matrix.createDenseFromNumpy(numpy_mat) - self.assertEqual((int(m.getHeight()), int(m.getWidth())), numpy_mat.shape) + self.assertEqual((int(m.getHeight()), int(m.getWidth())), + numpy_mat.shape) self.assertEqual(m.isGpu(), swig_paddle.isUsingGpu()) for a, e in zip(m.getData(), [1.0, 2.0, 3.0, 4.0, 5.0, 6.0]): self.assertAlmostEqual(a, e) diff --git a/paddle/api/test/testVector.py b/paddle/api/test/testVector.py index 525ed97edd..1ab095c1d3 100644 --- a/paddle/api/test/testVector.py +++ b/paddle/api/test/testVector.py @@ -26,17 +26,17 @@ class TestIVector(unittest.TestCase): self.assertEqual(m[i], 0) m[i] = i self.assertEqual(m[i], i) - + m = swig_paddle.IVector.createZero(10) self.assertEqual(m.isGpu(), swig_paddle.isUsingGpu()) - self.assertEqual(m.getData(), [0]*10) + self.assertEqual(m.getData(), [0] * 10) def test_create(self): m = swig_paddle.IVector.create(range(10), False) self.assertIsNotNone(m) for i in xrange(10): self.assertEqual(m[i], i) - + m = swig_paddle.IVector.create(range(10)) self.assertEqual(m.isGpu(), swig_paddle.isUsingGpu()) self.assertEqual(m.getData(), range(10)) @@ -69,7 +69,7 @@ class TestIVector(unittest.TestCase): expect_vec = range(0, 10) expect_vec[4] = 7 self.assertEqual(vec.getData(), expect_vec) - + def test_numpy(self): vec = np.array([1, 3, 4, 65, 78, 1, 4], dtype="int32") iv = swig_paddle.IVector.createVectorFromNumpy(vec) @@ -85,10 +85,10 @@ class TestVector(unittest.TestCase): self.assertTrue(util.doubleEqual(v[i], 0)) v[i] = i self.assertTrue(util.doubleEqual(v[i], i)) - + v = swig_paddle.Vector.createZero(10) self.assertEqual(v.isGpu(), swig_paddle.isUsingGpu()) - self.assertEqual(v.getData(), [0]*10) + self.assertEqual(v.getData(), [0] * 10) def testCreate(self): v = swig_paddle.Vector.create([x / 100.0 for x in xrange(100)], False) @@ -96,14 +96,13 @@ class TestVector(unittest.TestCase): for i in xrange(len(v)): self.assertTrue(util.doubleEqual(v[i], i / 100.0)) self.assertEqual(100, len(v)) - + v = swig_paddle.Vector.create([x / 100.0 for x in xrange(100)]) self.assertEqual(v.isGpu(), swig_paddle.isUsingGpu()) self.assertEqual(100, len(v)) vdata = v.getData() for i in xrange(len(v)): self.assertTrue(util.doubleEqual(vdata[i], i / 100.0)) - def testCpuNumpy(self): numpy_arr = np.array([1.2, 2.3, 3.4, 4.5], dtype="float32") @@ -128,7 +127,7 @@ class TestVector(unittest.TestCase): for i in xrange(1, len(numpy_3)): util.doubleEqual(numpy_3[i], vec[i]) - + def testNumpy(self): numpy_arr = np.array([1.2, 2.3, 3.4, 4.5], dtype="float32") vec = swig_paddle.Vector.createVectorFromNumpy(numpy_arr) @@ -136,7 +135,6 @@ class TestVector(unittest.TestCase): vecData = vec.getData() for n, v in zip(numpy_arr, vecData): self.assertTrue(util.doubleEqual(n, v)) - def testCopyFromNumpy(self): vec = swig_paddle.Vector.createZero(1, False) diff --git a/paddle/cuda/include/hl_base.h b/paddle/cuda/include/hl_base.h index 0b9dfc6117..84c5f2d5c9 100644 --- a/paddle/cuda/include/hl_base.h +++ b/paddle/cuda/include/hl_base.h @@ -223,9 +223,9 @@ typedef struct { #ifdef __NVCC__ -#include "paddle/utils/Logging.h" -#include "hl_cuda.h" #include "cuda_runtime.h" +#include "hl_cuda.h" +#include "paddle/utils/Logging.h" extern __thread bool g_sync_flag; extern __thread cudaStream_t default_stream; diff --git a/paddle/cuda/include/hl_dso_loader.h b/paddle/cuda/include/hl_dso_loader.h index 9ddf0e61ee..20c13f21e6 100644 --- a/paddle/cuda/include/hl_dso_loader.h +++ b/paddle/cuda/include/hl_dso_loader.h @@ -16,8 +16,8 @@ limitations under the License. */ #define HL_DSO_LOADER_H_ #include -#include #include +#include #include "hl_base.h" /** diff --git a/paddle/cuda/include/hl_gpu.h b/paddle/cuda/include/hl_gpu.h index aad0450c8c..ede2670882 100644 --- a/paddle/cuda/include/hl_gpu.h +++ b/paddle/cuda/include/hl_gpu.h @@ -15,28 +15,28 @@ limitations under the License. */ #ifndef HL_GPU_H_ #define HL_GPU_H_ +#include "hl_aggregate.h" #include "hl_base.h" +#include "hl_cnn.h" #include "hl_cuda.h" #include "hl_cuda_cublas.h" #include "hl_cuda_cudnn.h" -#include "hl_matrix.h" -#include "hl_aggregate.h" -#include "hl_cnn.h" -#include "hl_sparse.h" #include "hl_lstm.h" +#include "hl_matrix.h" #include "hl_sequence.h" +#include "hl_sparse.h" #include "hl_warpctc_wrap.h" #ifdef HPPL_STUB_FUNC -#include "stub/hl_cuda_stub.h" -#include "stub/hl_cuda_cublas_stub.h" -#include "stub/hl_cuda_cudnn_stub.h" -#include "stub/hl_matrix_stub.h" #include "stub/hl_aggregate_stub.h" #include "stub/hl_cnn_stub.h" -#include "stub/hl_sparse_stub.h" +#include "stub/hl_cuda_cublas_stub.h" +#include "stub/hl_cuda_cudnn_stub.h" +#include "stub/hl_cuda_stub.h" #include "stub/hl_lstm_stub.h" +#include "stub/hl_matrix_stub.h" #include "stub/hl_sequence_stub.h" +#include "stub/hl_sparse_stub.h" #endif #endif /* HL_GPU_H_ */ diff --git a/paddle/cuda/src/hl_cuda_cublas.cc b/paddle/cuda/src/hl_cuda_cublas.cc index 7cede8c63c..182e8ab218 100644 --- a/paddle/cuda/src/hl_cuda_cublas.cc +++ b/paddle/cuda/src/hl_cuda_cublas.cc @@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "hl_cuda_cublas.h" #include #include #include "hl_cuda.h" -#include "hl_cuda_cublas.h" -#include "hl_thread.ph" #include "hl_dso_loader.h" +#include "hl_thread.ph" #include "paddle/utils/Logging.h" namespace dynload { diff --git a/paddle/cuda/src/hl_cuda_cudnn.cc b/paddle/cuda/src/hl_cuda_cudnn.cc index 9c9b8906c2..7111224d59 100644 --- a/paddle/cuda/src/hl_cuda_cudnn.cc +++ b/paddle/cuda/src/hl_cuda_cudnn.cc @@ -12,14 +12,14 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "hl_cuda_cudnn.h" #include #include -#include "hl_cuda_cudnn.h" #include "hl_cuda_cudnn.ph" -#include "hl_thread.ph" #include "hl_dso_loader.h" -#include "paddle/utils/Logging.h" +#include "hl_thread.ph" #include "paddle/utils/CommandLineParser.h" +#include "paddle/utils/Logging.h" P_DEFINE_int32(cudnn_conv_workspace_limit_in_mb, 4096, diff --git a/paddle/cuda/src/hl_cudart_wrap.cc b/paddle/cuda/src/hl_cudart_wrap.cc index a3ac750b53..ecc03a729d 100644 --- a/paddle/cuda/src/hl_cudart_wrap.cc +++ b/paddle/cuda/src/hl_cudart_wrap.cc @@ -14,8 +14,8 @@ limitations under the License. */ #ifdef PADDLE_USE_DSO -#include #include +#include #include "hl_dso_loader.h" /** diff --git a/paddle/cuda/src/hl_time.cc b/paddle/cuda/src/hl_time.cc index 3005065899..2bb69d25e5 100644 --- a/paddle/cuda/src/hl_time.cc +++ b/paddle/cuda/src/hl_time.cc @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include +#include "hl_time.h" #include +#include #include -#include "hl_time.h" using std::chrono::high_resolution_clock; diff --git a/paddle/cuda/src/hl_warpctc_wrap.cc b/paddle/cuda/src/hl_warpctc_wrap.cc index 619b90120f..9ae8bc0f22 100644 --- a/paddle/cuda/src/hl_warpctc_wrap.cc +++ b/paddle/cuda/src/hl_warpctc_wrap.cc @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include "hl_warpctc_wrap.h" +#include #include "hl_dso_loader.h" #include "paddle/utils/Logging.h" diff --git a/paddle/gserver/activations/ActivationFunction.cpp b/paddle/gserver/activations/ActivationFunction.cpp index f1d09c568d..f8c4bcac2f 100644 --- a/paddle/gserver/activations/ActivationFunction.cpp +++ b/paddle/gserver/activations/ActivationFunction.cpp @@ -15,13 +15,13 @@ limitations under the License. */ #include "ActivationFunction.h" #include -#include #include -#include +#include #include #include -#include "paddle/utils/ClassRegistrar.h" +#include #include "paddle/parameter/Argument.h" +#include "paddle/utils/ClassRegistrar.h" #include "paddle/utils/Logging.h" diff --git a/paddle/gserver/dataproviders/DataProvider.cpp b/paddle/gserver/dataproviders/DataProvider.cpp index 55ca62543a..0478256f9c 100644 --- a/paddle/gserver/dataproviders/DataProvider.cpp +++ b/paddle/gserver/dataproviders/DataProvider.cpp @@ -14,12 +14,12 @@ limitations under the License. */ #include "DataProvider.h" -#include "paddle/utils/Util.h" -#include "paddle/utils/StringUtil.h" -#include "paddle/utils/Logging.h" -#include #include +#include #include "ProtoDataProvider.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/StringUtil.h" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/gserver/dataproviders/DataProvider.h b/paddle/gserver/dataproviders/DataProvider.h index 5b854936c6..9b7f7e36ce 100644 --- a/paddle/gserver/dataproviders/DataProvider.h +++ b/paddle/gserver/dataproviders/DataProvider.h @@ -14,28 +14,28 @@ limitations under the License. */ #pragma once -#include -#include -#include -#include -#include #include -#include -#include #include +#include +#include +#include +#include +#include +#include +#include +#include "DataConfig.pb.h" +#include "paddle/math/Matrix.h" +#include "paddle/math/SparseMatrix.h" +#include "paddle/math/Vector.h" +#include "paddle/parameter/Argument.h" +#include "paddle/utils/ClassRegistrar.h" +#include "paddle/utils/Locks.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Queue.h" -#include "paddle/utils/Locks.h" #include "paddle/utils/ThreadLocal.h" #include "paddle/utils/TypeDefs.h" -#include "paddle/math/Matrix.h" -#include "paddle/math/SparseMatrix.h" #include "paddle/utils/Util.h" -#include "paddle/math/Vector.h" -#include "DataConfig.pb.h" -#include "paddle/utils/ClassRegistrar.h" -#include "paddle/parameter/Argument.h" namespace paddle { /** diff --git a/paddle/gserver/dataproviders/MultiDataProvider.cpp b/paddle/gserver/dataproviders/MultiDataProvider.cpp index e1fc4c9365..46fe053768 100644 --- a/paddle/gserver/dataproviders/MultiDataProvider.cpp +++ b/paddle/gserver/dataproviders/MultiDataProvider.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" #include "MultiDataProvider.h" -#include "paddle/utils/Logging.h" #include +#include "paddle/utils/Logging.h" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/gserver/dataproviders/ProtoDataProvider.cpp b/paddle/gserver/dataproviders/ProtoDataProvider.cpp index 6a0cb5ef63..d16ecca2d9 100644 --- a/paddle/gserver/dataproviders/ProtoDataProvider.cpp +++ b/paddle/gserver/dataproviders/ProtoDataProvider.cpp @@ -13,14 +13,14 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "ProtoDataProvider.h" -#include "paddle/utils/Util.h" -#include "paddle/utils/StringUtil.h" #include #include #include +#include "paddle/utils/StringUtil.h" +#include "paddle/utils/Util.h" -#include "paddle/utils/Logging.h" #include "DataProviderGroup.h" +#include "paddle/utils/Logging.h" P_DEFINE_double(memory_threshold_on_load_data, 1.0, @@ -562,16 +562,16 @@ int64_t ProtoDataProvider::getNextBatchInternal(int64_t size, auto mat = cpuArguments[slot].value; mat->resize(size, dim); if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(dataPos.data(), - slots_[slot].indices.data(), - slots_[slot].sparseNonValueData.data(), - HPPL_STREAM_1); + std::dynamic_pointer_cast(mat)->copyFrom( + dataPos.data(), + slots_[slot].indices.data(), + slots_[slot].sparseNonValueData.data(), + HPPL_STREAM_1); } else if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(dataPos.data(), - slots_[slot].indices.data(), - slots_[slot].sparseNonValueData.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + dataPos.data(), + slots_[slot].indices.data(), + slots_[slot].sparseNonValueData.data()); } else { LOG(FATAL) << "Not Supported"; } @@ -598,16 +598,16 @@ int64_t ProtoDataProvider::getNextBatchInternal(int64_t size, auto mat = cpuArguments[slot].value; mat->resize(size, dim); if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(dataPos.data(), - slots_[slot].indices.data(), - slots_[slot].sparseFloatValueData.data(), - HPPL_STREAM_1); + std::dynamic_pointer_cast(mat)->copyFrom( + dataPos.data(), + slots_[slot].indices.data(), + slots_[slot].sparseFloatValueData.data(), + HPPL_STREAM_1); } else if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(dataPos.data(), - slots_[slot].indices.data(), - slots_[slot].sparseFloatValueData.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + dataPos.data(), + slots_[slot].indices.data(), + slots_[slot].sparseFloatValueData.data()); } else { LOG(FATAL) << "Not Supported"; } diff --git a/paddle/gserver/dataproviders/ProtoDataProvider.h b/paddle/gserver/dataproviders/ProtoDataProvider.h index 9ec5cb97c0..7dd45e0622 100644 --- a/paddle/gserver/dataproviders/ProtoDataProvider.h +++ b/paddle/gserver/dataproviders/ProtoDataProvider.h @@ -16,8 +16,8 @@ limitations under the License. */ #include -#include "paddle/utils/Stat.h" #include "DataFormat.pb.h" +#include "paddle/utils/Stat.h" #include "DataProvider.h" #include "ProtoReader.h" diff --git a/paddle/gserver/dataproviders/ProtoReader.h b/paddle/gserver/dataproviders/ProtoReader.h index 6708e7cde7..4e6f58a529 100644 --- a/paddle/gserver/dataproviders/ProtoReader.h +++ b/paddle/gserver/dataproviders/ProtoReader.h @@ -16,10 +16,10 @@ limitations under the License. */ #include -#include #include -#include #include +#include +#include namespace paddle { diff --git a/paddle/gserver/dataproviders/PyDataProvider.cpp b/paddle/gserver/dataproviders/PyDataProvider.cpp index f5dcbfcf34..5bdd55309c 100644 --- a/paddle/gserver/dataproviders/PyDataProvider.cpp +++ b/paddle/gserver/dataproviders/PyDataProvider.cpp @@ -13,10 +13,10 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "PyDataProvider.h" -#include "paddle/utils/PythonUtil.h" #include -#include "paddle/utils/Util.h" #include "paddle/utils/Excepts.h" +#include "paddle/utils/PythonUtil.h" +#include "paddle/utils/Util.h" namespace paddle { @@ -316,16 +316,16 @@ void PyDataProvider::handleSparseNonValueSlot( auto mat = cpuArguments[slotIndex].value; mat->resize(slot.sampleNum, dim, slot.sampleNum, NO_VALUE, SPARSE_CSR); if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(slot.sampleSequenceIdVec.data(), - slot.indices.data(), - slot.sparseNonValueData.data(), - HPPL_STREAM_1); + std::dynamic_pointer_cast(mat)->copyFrom( + slot.sampleSequenceIdVec.data(), + slot.indices.data(), + slot.sparseNonValueData.data(), + HPPL_STREAM_1); } else if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(slot.sampleSequenceIdVec.data(), - slot.indices.data(), - slot.sparseNonValueData.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + slot.sampleSequenceIdVec.data(), + slot.indices.data(), + slot.sparseNonValueData.data()); } else { LOG(FATAL) << "Not Supported"; } @@ -347,16 +347,16 @@ void PyDataProvider::handleSparseValueSlot( auto mat = cpuArguments[slotIndex].value; mat->resize(slot.sampleNum, dim, slot.sampleNum, FLOAT_VALUE, SPARSE_CSR); if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(slot.sampleSequenceIdVec.data(), - slot.indices.data(), - slot.sparseFloatValueData.data(), - HPPL_STREAM_DEFAULT); + std::dynamic_pointer_cast(mat)->copyFrom( + slot.sampleSequenceIdVec.data(), + slot.indices.data(), + slot.sparseFloatValueData.data(), + HPPL_STREAM_DEFAULT); } else if (std::dynamic_pointer_cast(mat)) { - std::dynamic_pointer_cast(mat) - ->copyFrom(slot.sampleSequenceIdVec.data(), - slot.indices.data(), - slot.sparseFloatValueData.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + slot.sampleSequenceIdVec.data(), + slot.indices.data(), + slot.sparseFloatValueData.data()); } else { LOG(FATAL) << "Not Supported"; } diff --git a/paddle/gserver/dataproviders/PyDataProvider2.cpp b/paddle/gserver/dataproviders/PyDataProvider2.cpp index 8b04a03f6d..460efc5adc 100644 --- a/paddle/gserver/dataproviders/PyDataProvider2.cpp +++ b/paddle/gserver/dataproviders/PyDataProvider2.cpp @@ -15,18 +15,18 @@ limitations under the License. */ #ifndef PADDLE_NO_PYTHON #include +#include #include #include -#include #include -#include +#include #define NPY_NO_DEPRECATED_API NPY_1_7_API_VERSION #include #include "DataProvider.h" -#include "paddle/utils/PythonUtil.h" #include "paddle/utils/Locks.h" +#include "paddle/utils/PythonUtil.h" #include "paddle/utils/Stat.h" namespace paddle { @@ -400,10 +400,9 @@ private: if (this->loadThread_) { // wait poolActualSize < poolSize; std::unique_lock l(mtx_); - pushCV_.wait(l, - [this, additionalBatchSize] { - return this->poolActualSize_ < poolSize_; - }); + pushCV_.wait(l, [this, additionalBatchSize] { + return this->poolActualSize_ < poolSize_; + }); } { @@ -529,12 +528,10 @@ public: // but, loading from cache, cache object should ensure // data pool ready. std::unique_lock l(mtx_); - pullCV_.wait(l, - [this, &size] { - return this->poolActualSize_ >= - std::max(size, this->minPoolSize_) || - callingContexts_.empty(); - }); + pullCV_.wait(l, [this, &size] { + return this->poolActualSize_ >= std::max(size, this->minPoolSize_) || + callingContexts_.empty(); + }); if (unittest::OnPoolFilled) { (*unittest::OnPoolFilled)(this->poolActualSize_); diff --git a/paddle/gserver/evaluators/Evaluator.cpp b/paddle/gserver/evaluators/Evaluator.cpp index aa6dc7cb86..7556d21e01 100644 --- a/paddle/gserver/evaluators/Evaluator.cpp +++ b/paddle/gserver/evaluators/Evaluator.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "paddle/gserver/evaluators/Evaluator.h" +#include "paddle/utils/Stat.h" #include "paddle/gserver/gradientmachines/NeuralNetwork.h" @@ -842,9 +842,9 @@ void PnpairEvaluator::calc(std::vector& predictArray) { auto start = predictArray.begin(); while (start != predictArray.end()) { auto end = std::find_if( - start + 1, - predictArray.end(), - [=](const PredictionResult& x) { return x.queryid != start->queryid; }); + start + 1, predictArray.end(), [=](const PredictionResult& x) { + return x.queryid != start->queryid; + }); CHECK(end != start); stat(start - predictArray.begin(), end - predictArray.begin(), diff --git a/paddle/gserver/evaluators/Evaluator.h b/paddle/gserver/evaluators/Evaluator.h index a26c650c38..5770847309 100644 --- a/paddle/gserver/evaluators/Evaluator.h +++ b/paddle/gserver/evaluators/Evaluator.h @@ -14,11 +14,11 @@ limitations under the License. */ #pragma once -#include "paddle/pserver/ParameterClient2.h" -#include "paddle/utils/ClassRegistrar.h" +#include #include "ModelConfig.pb.h" #include "paddle/parameter/Argument.h" -#include +#include "paddle/pserver/ParameterClient2.h" +#include "paddle/utils/ClassRegistrar.h" namespace paddle { diff --git a/paddle/gserver/gradientmachines/GradientMachine.cpp b/paddle/gserver/gradientmachines/GradientMachine.cpp index 6adee05dbe..36ca05b919 100644 --- a/paddle/gserver/gradientmachines/GradientMachine.cpp +++ b/paddle/gserver/gradientmachines/GradientMachine.cpp @@ -14,16 +14,16 @@ limitations under the License. */ #include "GradientMachine.h" -#include "paddle/utils/Logging.h" #include +#include "paddle/utils/Logging.h" -#include "hl_gpu.h" -#include "NeuralNetwork.h" -#include "ParallelNeuralNetwork.h" +#include "GradientMachineMode.h" #include "MultiGradientMachine.h" -#include "NeuralNetwork.h" #include "MultiNetwork.h" -#include "GradientMachineMode.h" +#include "NeuralNetwork.h" +#include "NeuralNetwork.h" +#include "ParallelNeuralNetwork.h" +#include "hl_gpu.h" namespace paddle { diff --git a/paddle/gserver/gradientmachines/GradientMachine.h b/paddle/gserver/gradientmachines/GradientMachine.h index f3e44a9e39..579eca71d4 100644 --- a/paddle/gserver/gradientmachines/GradientMachine.h +++ b/paddle/gserver/gradientmachines/GradientMachine.h @@ -17,15 +17,15 @@ limitations under the License. */ #include #include -#include "paddle/math/Matrix.h" -#include "paddle/parameter/Parameter.h" -#include "paddle/parameter/ParameterUpdaterBase.h" -#include "paddle/utils/Thread.h" -#include "TrainerConfig.pb.h" #include "ModelConfig.pb.h" +#include "TrainerConfig.pb.h" #include "paddle/gserver/dataproviders/DataProvider.h" #include "paddle/gserver/evaluators/Evaluator.h" #include "paddle/gserver/layers/Layer.h" +#include "paddle/math/Matrix.h" +#include "paddle/parameter/Parameter.h" +#include "paddle/parameter/ParameterUpdaterBase.h" +#include "paddle/utils/Thread.h" namespace paddle { /** diff --git a/paddle/gserver/gradientmachines/MultiGradientMachine.h b/paddle/gserver/gradientmachines/MultiGradientMachine.h index fe6d96e8ea..5f9855c4be 100644 --- a/paddle/gserver/gradientmachines/MultiGradientMachine.h +++ b/paddle/gserver/gradientmachines/MultiGradientMachine.h @@ -18,9 +18,9 @@ limitations under the License. */ #include "GradientMachine.h" -#include "paddle/utils/Queue.h" -#include "paddle/utils/Locks.h" #include "hl_gpu.h" +#include "paddle/utils/Locks.h" +#include "paddle/utils/Queue.h" namespace paddle { diff --git a/paddle/gserver/gradientmachines/MultiNetwork.cpp b/paddle/gserver/gradientmachines/MultiNetwork.cpp index 61af82fcb7..6eb3d8db96 100644 --- a/paddle/gserver/gradientmachines/MultiNetwork.cpp +++ b/paddle/gserver/gradientmachines/MultiNetwork.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include #include "paddle/utils/Stat.h" #include "paddle/utils/Util.h" -#include #include "MultiNetwork.h" diff --git a/paddle/gserver/gradientmachines/NeuralNetwork.cpp b/paddle/gserver/gradientmachines/NeuralNetwork.cpp index dbcb97b42b..ee36a87b9d 100644 --- a/paddle/gserver/gradientmachines/NeuralNetwork.cpp +++ b/paddle/gserver/gradientmachines/NeuralNetwork.cpp @@ -14,15 +14,15 @@ limitations under the License. */ #include "paddle/utils/Util.h" -#include "paddle/utils/Logging.h" #include "paddle/utils/CustomStackTrace.h" +#include "paddle/utils/Logging.h" -#include "paddle/utils/Stat.h" -#include "hl_gpu.h" +#include "MultiNetwork.h" #include "NeuralNetwork.h" #include "RecurrentGradientMachine.h" -#include "MultiNetwork.h" +#include "hl_gpu.h" #include "paddle/gserver/layers/AgentLayer.h" +#include "paddle/utils/Stat.h" namespace paddle { void parameterInitNN(int paramId, diff --git a/paddle/gserver/gradientmachines/NeuralNetwork.h b/paddle/gserver/gradientmachines/NeuralNetwork.h index fd885b436a..384ca88f47 100644 --- a/paddle/gserver/gradientmachines/NeuralNetwork.h +++ b/paddle/gserver/gradientmachines/NeuralNetwork.h @@ -14,18 +14,18 @@ limitations under the License. */ #pragma once -#include -#include #include +#include +#include -#include "paddle/utils/ClassRegistrar.h" -#include "paddle/parameter/Parameter.h" #include "ModelConfig.pb.h" +#include "paddle/gserver/dataproviders/DataProvider.h" #include "paddle/gserver/gradientmachines/GradientMachine.h" #include "paddle/gserver/layers/CostLayer.h" #include "paddle/gserver/layers/DataLayer.h" -#include "paddle/gserver/dataproviders/DataProvider.h" #include "paddle/gserver/layers/Layer.h" +#include "paddle/parameter/Parameter.h" +#include "paddle/utils/ClassRegistrar.h" namespace paddle { /* @@ -57,14 +57,13 @@ void parameterInitNN(int paramId, class NeuralNetwork : public GradientMachine { public: - virtual void init( - const ModelConfig& config, - ParamInitCallback callback = nullptr, - const std::vector& - parameterTypes = std::vector{PARAMETER_VALUE, - PARAMETER_GRADIENT, - PARAMETER_MOMENTUM}, - bool useGpu = FLAGS_use_gpu); + virtual void init(const ModelConfig& config, + ParamInitCallback callback = nullptr, + const std::vector& parameterTypes = + std::vector{PARAMETER_VALUE, + PARAMETER_GRADIENT, + PARAMETER_MOMENTUM}, + bool useGpu = FLAGS_use_gpu); /** * Connect two submodels and diff --git a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h index 934a7cfc7b..8f445b1ded 100644 --- a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h +++ b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h @@ -37,14 +37,13 @@ public: NeuralNetwork *rootNetwork = nullptr) : NeuralNetwork(subModelName, rootNetwork) {} - virtual void init( - const ModelConfig &config, - ParamInitCallback callback = nullptr, - const std::vector - ¶meterTypes = std::vector{PARAMETER_VALUE, - PARAMETER_GRADIENT, - PARAMETER_MOMENTUM}, - bool useGpu = FLAGS_use_gpu); + virtual void init(const ModelConfig &config, + ParamInitCallback callback = nullptr, + const std::vector ¶meterTypes = + std::vector{PARAMETER_VALUE, + PARAMETER_GRADIENT, + PARAMETER_MOMENTUM}, + bool useGpu = FLAGS_use_gpu); virtual void forward(const std::vector &inArgs, std::vector *outArgs, diff --git a/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp b/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp index 4fb1a44ab7..ee1c92bdf5 100644 --- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp +++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp @@ -12,17 +12,17 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" -#include "paddle/utils/Util.h" -#include "paddle/utils/Flags.h" +#include "RecurrentGradientMachine.h" +#include #include +#include #include -#include #include -#include -#include "RecurrentGradientMachine.h" #include "NeuralNetwork.h" #include "paddle/gserver/layers/AgentLayer.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Stat.h" +#include "paddle/utils/Util.h" P_DEFINE_string(diy_beam_search_prob_so, "", "the diy beam search cost so"); @@ -78,20 +78,22 @@ static inline SymbolType loadDiySymbol(const char* symbolName) { return reinterpret_cast(sym); } -static InitFunction __init__diy_prob_method([] { - std::string soName = FLAGS_diy_beam_search_prob_so; - if (!soName.empty()) { - gDiyProbHandle = dlopen(soName.c_str(), RTLD_LAZY); - CHECK(gDiyProbHandle) << "Cannot Open DIY Prob So " << soName; - atexit(exit_diy_prob); - gDiyProbMethod = - loadDiySymbol(DIY_CALC_PROB_SYMBOL_NAME); - gDiyProbStart = - loadDiySymbol(DIY_START_CALC_PROB_SYMBOL_NAME); - gDiyProbStop = - loadDiySymbol(DIY_FINISH_CALC_PROB_SYMBOL_NAME); - } -}, std::numeric_limits::max()); +static InitFunction __init__diy_prob_method( + [] { + std::string soName = FLAGS_diy_beam_search_prob_so; + if (!soName.empty()) { + gDiyProbHandle = dlopen(soName.c_str(), RTLD_LAZY); + CHECK(gDiyProbHandle) << "Cannot Open DIY Prob So " << soName; + atexit(exit_diy_prob); + gDiyProbMethod = + loadDiySymbol(DIY_CALC_PROB_SYMBOL_NAME); + gDiyProbStart = loadDiySymbol( + DIY_START_CALC_PROB_SYMBOL_NAME); + gDiyProbStop = loadDiySymbol( + DIY_FINISH_CALC_PROB_SYMBOL_NAME); + } + }, + std::numeric_limits::max()); class BeamSearchControlCallbacks { public: @@ -1281,10 +1283,9 @@ void RecurrentGradientMachine::beamSearch(size_t batchSize) { std::vector*> prefixes; prefixes.resize(paths.size()); std::transform( - paths.begin(), - paths.end(), - prefixes.begin(), - [](const Path& p) { return const_cast*>(&p.ids); }); + paths.begin(), paths.end(), prefixes.begin(), [](const Path& p) { + return const_cast*>(&p.ids); + }); beamSearchCtrlCallbacks_->beamSearchCandidateAdjust( prefixes, frames_[machineCur].get(), i); } diff --git a/paddle/gserver/gradientmachines/RecurrentGradientMachine.h b/paddle/gserver/gradientmachines/RecurrentGradientMachine.h index 369c8c3d98..db7d8aff6d 100644 --- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.h +++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once +#include #include "GradientMachine.h" #include "NeuralNetwork.h" -#include #include "paddle/utils/Locks.h" diff --git a/paddle/gserver/layers/BatchNormBaseLayer.cpp b/paddle/gserver/layers/BatchNormBaseLayer.cpp index 51463f1118..1ceaaaa206 100644 --- a/paddle/gserver/layers/BatchNormBaseLayer.cpp +++ b/paddle/gserver/layers/BatchNormBaseLayer.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" -#include "Layer.h" #include "BatchNormBaseLayer.h" #include "BatchNormalizationLayer.h" +#include "Layer.h" +#include "paddle/utils/Stat.h" #ifndef PADDLE_ONLY_CPU #include "CudnnBatchNormLayer.h" #endif diff --git a/paddle/gserver/layers/BatchNormBaseLayer.h b/paddle/gserver/layers/BatchNormBaseLayer.h index f5a555a6d0..75bda95de1 100644 --- a/paddle/gserver/layers/BatchNormBaseLayer.h +++ b/paddle/gserver/layers/BatchNormBaseLayer.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include "paddle/utils/Stat.h" #include "Layer.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/BatchNormalizationLayer.h b/paddle/gserver/layers/BatchNormalizationLayer.h index 56be473568..052c207732 100644 --- a/paddle/gserver/layers/BatchNormalizationLayer.h +++ b/paddle/gserver/layers/BatchNormalizationLayer.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include "Layer.h" #include "BatchNormBaseLayer.h" +#include "Layer.h" namespace paddle { diff --git a/paddle/gserver/layers/ConcatenateLayer.cpp b/paddle/gserver/layers/ConcatenateLayer.cpp index f6b3d86b8c..d19adace7d 100644 --- a/paddle/gserver/layers/ConcatenateLayer.cpp +++ b/paddle/gserver/layers/ConcatenateLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "Layer.h" #include "Projection.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/ContextProjection.cpp b/paddle/gserver/layers/ContextProjection.cpp index 6080aa51b9..7ac56e3a2a 100644 --- a/paddle/gserver/layers/ContextProjection.cpp +++ b/paddle/gserver/layers/ContextProjection.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "ContextProjection.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/ConvBaseLayer.cpp b/paddle/gserver/layers/ConvBaseLayer.cpp index 473ca24a94..7b234dc2a6 100644 --- a/paddle/gserver/layers/ConvBaseLayer.cpp +++ b/paddle/gserver/layers/ConvBaseLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "ConvBaseLayer.h" #include "paddle/math/MathUtils.h" +#include "paddle/utils/Logging.h" namespace paddle { bool ConvBaseLayer::init(const LayerMap& layerMap, diff --git a/paddle/gserver/layers/ConvOperator.cpp b/paddle/gserver/layers/ConvOperator.cpp index 3ede98ba4b..f943410dee 100644 --- a/paddle/gserver/layers/ConvOperator.cpp +++ b/paddle/gserver/layers/ConvOperator.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/math/Matrix.h" -#include "paddle/math/MathUtils.h" #include "Operator.h" +#include "paddle/math/MathUtils.h" +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/ConvProjection.cpp b/paddle/gserver/layers/ConvProjection.cpp index e72dc37ec8..aa634b3287 100644 --- a/paddle/gserver/layers/ConvProjection.cpp +++ b/paddle/gserver/layers/ConvProjection.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "ConvProjection.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/ConvShiftLayer.cpp b/paddle/gserver/layers/ConvShiftLayer.cpp index 527d885d86..9bfb1ab7a4 100644 --- a/paddle/gserver/layers/ConvShiftLayer.cpp +++ b/paddle/gserver/layers/ConvShiftLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/ConvexCombinationLayer.cpp b/paddle/gserver/layers/ConvexCombinationLayer.cpp index 57ff95fe37..3f4d77a2fe 100644 --- a/paddle/gserver/layers/ConvexCombinationLayer.cpp +++ b/paddle/gserver/layers/ConvexCombinationLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/CosSimVecMatLayer.cpp b/paddle/gserver/layers/CosSimVecMatLayer.cpp index e8a7f671ee..ad490b0b8c 100644 --- a/paddle/gserver/layers/CosSimVecMatLayer.cpp +++ b/paddle/gserver/layers/CosSimVecMatLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/CostLayer.cpp b/paddle/gserver/layers/CostLayer.cpp index 90cd473c42..7e9519f6b3 100644 --- a/paddle/gserver/layers/CostLayer.cpp +++ b/paddle/gserver/layers/CostLayer.cpp @@ -12,11 +12,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include +#include "CostLayer.h" #include -#include "paddle/utils/Logging.h" #include -#include "CostLayer.h" +#include +#include "paddle/utils/Logging.h" #include "paddle/math/SparseMatrix.h" diff --git a/paddle/gserver/layers/CudnnBatchNormLayer.cpp b/paddle/gserver/layers/CudnnBatchNormLayer.cpp index d44c217105..09dac05a7a 100644 --- a/paddle/gserver/layers/CudnnBatchNormLayer.cpp +++ b/paddle/gserver/layers/CudnnBatchNormLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" -#include "Layer.h" #include "CudnnBatchNormLayer.h" +#include "Layer.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/CudnnBatchNormLayer.h b/paddle/gserver/layers/CudnnBatchNormLayer.h index a52a683e15..b1e7d2082f 100644 --- a/paddle/gserver/layers/CudnnBatchNormLayer.h +++ b/paddle/gserver/layers/CudnnBatchNormLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include "paddle/utils/Stat.h" -#include "Layer.h" #include "BatchNormBaseLayer.h" +#include "Layer.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/CudnnConvLayer.cpp b/paddle/gserver/layers/CudnnConvLayer.cpp index 6e28d5eb42..978c2c1479 100644 --- a/paddle/gserver/layers/CudnnConvLayer.cpp +++ b/paddle/gserver/layers/CudnnConvLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "CudnnConvLayer.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "CudnnConvLayer.h" namespace paddle { diff --git a/paddle/gserver/layers/CudnnConvLayer.h b/paddle/gserver/layers/CudnnConvLayer.h index 6317fab6f8..b869c695bd 100644 --- a/paddle/gserver/layers/CudnnConvLayer.h +++ b/paddle/gserver/layers/CudnnConvLayer.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once +#include #include "ConvBaseLayer.h" -#include "paddle/math/Matrix.h" #include "Projection.h" -#include +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/CudnnPoolLayer.cpp b/paddle/gserver/layers/CudnnPoolLayer.cpp index d0e71c6345..4adb2d4709 100644 --- a/paddle/gserver/layers/CudnnPoolLayer.cpp +++ b/paddle/gserver/layers/CudnnPoolLayer.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "CudnnPoolLayer.h" +#include "paddle/math/Matrix.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "paddle/math/Matrix.h" -#include "CudnnPoolLayer.h" namespace paddle { diff --git a/paddle/gserver/layers/EosIdCheckLayer.cpp b/paddle/gserver/layers/EosIdCheckLayer.cpp index dc3c6e6b64..fa53e2e4cf 100644 --- a/paddle/gserver/layers/EosIdCheckLayer.cpp +++ b/paddle/gserver/layers/EosIdCheckLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" +#include "paddle/utils/Logging.h" namespace paddle { /** diff --git a/paddle/gserver/layers/ExpandConvBaseLayer.h b/paddle/gserver/layers/ExpandConvBaseLayer.h index e14f6e6f44..8445642217 100644 --- a/paddle/gserver/layers/ExpandConvBaseLayer.h +++ b/paddle/gserver/layers/ExpandConvBaseLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once +#include #include "ConvBaseLayer.h" #include "paddle/math/Matrix.h" -#include namespace paddle { diff --git a/paddle/gserver/layers/ExpandConvLayer.cpp b/paddle/gserver/layers/ExpandConvLayer.cpp index dcc7839960..f9267b81a7 100644 --- a/paddle/gserver/layers/ExpandConvLayer.cpp +++ b/paddle/gserver/layers/ExpandConvLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "ExpandConvLayer.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "ExpandConvLayer.h" namespace paddle { diff --git a/paddle/gserver/layers/ExpandConvLayer.h b/paddle/gserver/layers/ExpandConvLayer.h index 6f8504b50a..de81a017e1 100644 --- a/paddle/gserver/layers/ExpandConvLayer.h +++ b/paddle/gserver/layers/ExpandConvLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include "paddle/math/Matrix.h" #include #include "ExpandConvBaseLayer.h" +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/ExpandConvTransLayer.cpp b/paddle/gserver/layers/ExpandConvTransLayer.cpp index cd4965c3c5..520586b138 100644 --- a/paddle/gserver/layers/ExpandConvTransLayer.cpp +++ b/paddle/gserver/layers/ExpandConvTransLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "ExpandConvTransLayer.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "ExpandConvTransLayer.h" /* The implementation of the convTransLayer is basically a swap of forward and * backward of the original convLayer. diff --git a/paddle/gserver/layers/ExpandConvTransLayer.h b/paddle/gserver/layers/ExpandConvTransLayer.h index fa9d7fb481..4a527d6799 100644 --- a/paddle/gserver/layers/ExpandConvTransLayer.h +++ b/paddle/gserver/layers/ExpandConvTransLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include "paddle/math/Matrix.h" #include #include "ExpandConvBaseLayer.h" +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/FullyConnectedLayer.cpp b/paddle/gserver/layers/FullyConnectedLayer.cpp index d2a028dd80..89afe33c36 100644 --- a/paddle/gserver/layers/FullyConnectedLayer.cpp +++ b/paddle/gserver/layers/FullyConnectedLayer.cpp @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "FullyConnectedLayer.h" +#include +#include +#include "paddle/math/SparseMatrix.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "paddle/math/SparseMatrix.h" -#include -#include namespace paddle { diff --git a/paddle/gserver/layers/GatedRecurrentLayer.cpp b/paddle/gserver/layers/GatedRecurrentLayer.cpp index 01b210ba70..930d9a0561 100644 --- a/paddle/gserver/layers/GatedRecurrentLayer.cpp +++ b/paddle/gserver/layers/GatedRecurrentLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "Layer.h" #include "GatedRecurrentLayer.h" +#include "Layer.h" #include "paddle/utils/Stat.h" namespace paddle { @@ -386,8 +386,9 @@ void GatedRecurrentLayer::backwardBatch(int batchSize, MatrixPtr inputGrad) { { batchSize = outputGradTmp->getHeight(); gruValue.prevOutValue = - (n == 0 ? nullptr : (batchValue_->getBatchValue(n - 1, batchSize)) - ->getData()); + (n == 0 + ? nullptr + : (batchValue_->getBatchValue(n - 1, batchSize))->getData()); gruGrad.prevOutGrad = (n == 0 ? nullptr : (batchGrad_->getBatchValue(n - 1, batchSize))->getData()); diff --git a/paddle/gserver/layers/GatedRecurrentLayer.h b/paddle/gserver/layers/GatedRecurrentLayer.h index e099b4d18b..25770ce57f 100644 --- a/paddle/gserver/layers/GatedRecurrentLayer.h +++ b/paddle/gserver/layers/GatedRecurrentLayer.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once -#include "paddle/math/Matrix.h" -#include "SequenceToBatch.h" #include "GruCompute.h" #include "Layer.h" +#include "SequenceToBatch.h" +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/GruCompute.cpp b/paddle/gserver/layers/GruCompute.cpp index 7d4e8001a8..06907768e9 100644 --- a/paddle/gserver/layers/GruCompute.cpp +++ b/paddle/gserver/layers/GruCompute.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" #include "GruCompute.h" #include "hl_recurrent_apply.cuh" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/gserver/layers/GruCompute.h b/paddle/gserver/layers/GruCompute.h index 2a5da72068..42c0019319 100644 --- a/paddle/gserver/layers/GruCompute.h +++ b/paddle/gserver/layers/GruCompute.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include "paddle/utils/TypeDefs.h" #include "ModelConfig.pb.h" #include "hl_gpu.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/gserver/layers/GruStepLayer.cpp b/paddle/gserver/layers/GruStepLayer.cpp index c48b5e40e6..4a1006aa94 100644 --- a/paddle/gserver/layers/GruStepLayer.cpp +++ b/paddle/gserver/layers/GruStepLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "Layer.h" #include "GruCompute.h" +#include "Layer.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/IdentityProjection.cpp b/paddle/gserver/layers/IdentityProjection.cpp index 8660631b5a..f1d41a33d4 100644 --- a/paddle/gserver/layers/IdentityProjection.cpp +++ b/paddle/gserver/layers/IdentityProjection.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "Projection.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/InterpolationLayer.cpp b/paddle/gserver/layers/InterpolationLayer.cpp index 94d4614b21..44fe1fb1fe 100644 --- a/paddle/gserver/layers/InterpolationLayer.cpp +++ b/paddle/gserver/layers/InterpolationLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/Layer.cpp b/paddle/gserver/layers/Layer.cpp index 3c539f3076..c9e121047b 100644 --- a/paddle/gserver/layers/Layer.cpp +++ b/paddle/gserver/layers/Layer.cpp @@ -14,15 +14,15 @@ limitations under the License. */ #include "paddle/utils/Util.h" -#include "paddle/utils/Logging.h" #include "paddle/math/SparseMatrix.h" +#include "paddle/utils/Logging.h" #include "AddtoLayer.h" +#include "CRFLayer.h" #include "CosSimLayer.h" #include "CostLayer.h" -#include "ExpandConvLayer.h" -#include "CRFLayer.h" #include "DataLayer.h" +#include "ExpandConvLayer.h" #include "FullyConnectedLayer.h" #include "HierarchicalSigmoidLayer.h" #include "MaxLayer.h" diff --git a/paddle/gserver/layers/Layer.h b/paddle/gserver/layers/Layer.h index 6609e16c4c..172e558b82 100644 --- a/paddle/gserver/layers/Layer.h +++ b/paddle/gserver/layers/Layer.h @@ -14,18 +14,18 @@ limitations under the License. */ #pragma once -#include -#include #include -#include "paddle/utils/ClassRegistrar.h" +#include +#include +#include "ModelConfig.pb.h" #include "paddle/math/CpuSparseMatrix.h" #include "paddle/parameter/Parameter.h" +#include "paddle/utils/ClassRegistrar.h" #include "paddle/utils/Util.h" -#include "ModelConfig.pb.h" -#include "paddle/gserver/activations/ActivationFunction.h" #include #include +#include "paddle/gserver/activations/ActivationFunction.h" /// Macro for registering a layer type. /// Example: REGISTER_LAYER(crf_error, CRFDecodingErrorLayer); diff --git a/paddle/gserver/layers/LinearChainCRF.cpp b/paddle/gserver/layers/LinearChainCRF.cpp index c6414c822e..af550c7a01 100644 --- a/paddle/gserver/layers/LinearChainCRF.cpp +++ b/paddle/gserver/layers/LinearChainCRF.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include "LinearChainCRF.h" +#include namespace paddle { diff --git a/paddle/gserver/layers/LinearChainCTC.cpp b/paddle/gserver/layers/LinearChainCTC.cpp index 60e814fc30..cb2b249110 100644 --- a/paddle/gserver/layers/LinearChainCTC.cpp +++ b/paddle/gserver/layers/LinearChainCTC.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include "LinearChainCTC.h" +#include #include namespace paddle { diff --git a/paddle/gserver/layers/LstmCompute.cpp b/paddle/gserver/layers/LstmCompute.cpp index 18f7996958..4c42970964 100644 --- a/paddle/gserver/layers/LstmCompute.cpp +++ b/paddle/gserver/layers/LstmCompute.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" -#include "hl_recurrent_apply.cuh" #include "LstmCompute.h" +#include "hl_recurrent_apply.cuh" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/gserver/layers/LstmCompute.h b/paddle/gserver/layers/LstmCompute.h index 9b7aee19dd..140a4c6ecf 100644 --- a/paddle/gserver/layers/LstmCompute.h +++ b/paddle/gserver/layers/LstmCompute.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include "paddle/utils/TypeDefs.h" #include "ModelConfig.pb.h" #include "hl_gpu.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/gserver/layers/LstmLayer.cpp b/paddle/gserver/layers/LstmLayer.cpp index 975edcfe7f..452091eff4 100644 --- a/paddle/gserver/layers/LstmLayer.cpp +++ b/paddle/gserver/layers/LstmLayer.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "LstmLayer.h" -#include "paddle/math/Matrix.h" #include "paddle/math/BaseMatrix.h" +#include "paddle/math/Matrix.h" #include "paddle/utils/Stat.h" P_DECLARE_bool(prev_batch_state); diff --git a/paddle/gserver/layers/LstmLayer.h b/paddle/gserver/layers/LstmLayer.h index 16c62aa88d..f49df2c412 100644 --- a/paddle/gserver/layers/LstmLayer.h +++ b/paddle/gserver/layers/LstmLayer.h @@ -15,10 +15,10 @@ limitations under the License. */ #pragma once #include "Layer.h" -#include "paddle/math/Matrix.h" -#include "paddle/math/BaseMatrix.h" -#include "SequenceToBatch.h" #include "LstmCompute.h" +#include "SequenceToBatch.h" +#include "paddle/math/BaseMatrix.h" +#include "paddle/math/Matrix.h" namespace paddle { /** diff --git a/paddle/gserver/layers/MDLstmLayer.cpp b/paddle/gserver/layers/MDLstmLayer.cpp index 9d3797d16f..1243c12889 100644 --- a/paddle/gserver/layers/MDLstmLayer.cpp +++ b/paddle/gserver/layers/MDLstmLayer.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "LstmLayer.h" -#include "paddle/math/Matrix.h" #include "paddle/math/BaseMatrix.h" +#include "paddle/math/Matrix.h" namespace paddle { @@ -318,7 +318,7 @@ void MDLstmLayer::forward(PassType passType) { CHECK_EQ(starts[numSequences], batchSize); int* dimsData = input.cpuSequenceDims->getData(); - CHECK_EQ(int(input.cpuSequenceDims->getSize()), numDims_ * numSequences); + CHECK_EQ(int(input.cpuSequenceDims->getSize()), numDims_* numSequences); for (int i = 0; i < numSequences; i++) { std::vector dims; diff --git a/paddle/gserver/layers/MaxOutLayer.cpp b/paddle/gserver/layers/MaxOutLayer.cpp index 4fb99ce2a2..3a86a95321 100644 --- a/paddle/gserver/layers/MaxOutLayer.cpp +++ b/paddle/gserver/layers/MaxOutLayer.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "MaxOutLayer.h" -#include "hl_gpu.h" #include "hl_cnn.h" +#include "hl_gpu.h" namespace paddle { diff --git a/paddle/gserver/layers/MixedLayer.cpp b/paddle/gserver/layers/MixedLayer.cpp index 490b217347..2525b1984b 100644 --- a/paddle/gserver/layers/MixedLayer.cpp +++ b/paddle/gserver/layers/MixedLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "MixedLayer.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/MixedLayer.h b/paddle/gserver/layers/MixedLayer.h index d73ba6b7a1..9655a152c7 100644 --- a/paddle/gserver/layers/MixedLayer.h +++ b/paddle/gserver/layers/MixedLayer.h @@ -15,8 +15,8 @@ limitations under the License. */ #pragma once #include "Layer.h" -#include "Projection.h" #include "Operator.h" +#include "Projection.h" namespace paddle { diff --git a/paddle/gserver/layers/MultiplexLayer.cpp b/paddle/gserver/layers/MultiplexLayer.cpp index dc4a1ec321..d09720c525 100644 --- a/paddle/gserver/layers/MultiplexLayer.cpp +++ b/paddle/gserver/layers/MultiplexLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/NormLayer.cpp b/paddle/gserver/layers/NormLayer.cpp index b8682a1422..3db0af2515 100644 --- a/paddle/gserver/layers/NormLayer.cpp +++ b/paddle/gserver/layers/NormLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "NormLayer.h" #include "NormProjectionLayer.h" +#include "paddle/utils/Logging.h" namespace paddle { REGISTER_LAYER_CREATE_FUNC(norm, &NormLayer::create); diff --git a/paddle/gserver/layers/NormLayer.h b/paddle/gserver/layers/NormLayer.h index aedbb95b4f..86255b231b 100644 --- a/paddle/gserver/layers/NormLayer.h +++ b/paddle/gserver/layers/NormLayer.h @@ -16,8 +16,8 @@ limitations under the License. */ #include #include "Layer.h" -#include "paddle/math/Matrix.h" #include "NormLayer.h" +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/NormProjectionLayer.cpp b/paddle/gserver/layers/NormProjectionLayer.cpp index ea301292e0..934fc31e0a 100644 --- a/paddle/gserver/layers/NormProjectionLayer.cpp +++ b/paddle/gserver/layers/NormProjectionLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "NormProjectionLayer.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "NormProjectionLayer.h" namespace paddle { size_t CMRProjectionNormLayer::getSize() { diff --git a/paddle/gserver/layers/NormProjectionLayer.h b/paddle/gserver/layers/NormProjectionLayer.h index 0db8e2551f..4f7b638334 100644 --- a/paddle/gserver/layers/NormProjectionLayer.h +++ b/paddle/gserver/layers/NormProjectionLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once +#include #include "NormLayer.h" #include "paddle/math/Matrix.h" -#include namespace paddle { diff --git a/paddle/gserver/layers/Operator.h b/paddle/gserver/layers/Operator.h index b0586b59e9..6fd331382f 100644 --- a/paddle/gserver/layers/Operator.h +++ b/paddle/gserver/layers/Operator.h @@ -14,11 +14,11 @@ limitations under the License. */ #pragma once -#include "paddle/parameter/Parameter.h" #include "ModelConfig.pb.h" +#include "paddle/parameter/Parameter.h" -#include "paddle/parameter/Argument.h" #include "Layer.h" +#include "paddle/parameter/Argument.h" namespace paddle { diff --git a/paddle/gserver/layers/OuterProdLayer.cpp b/paddle/gserver/layers/OuterProdLayer.cpp index 42587dcce5..cf9a008318 100644 --- a/paddle/gserver/layers/OuterProdLayer.cpp +++ b/paddle/gserver/layers/OuterProdLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/PoolLayer.cpp b/paddle/gserver/layers/PoolLayer.cpp index 36e396487e..96d5c54acc 100644 --- a/paddle/gserver/layers/PoolLayer.cpp +++ b/paddle/gserver/layers/PoolLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "PoolLayer.h" #include "PoolProjectionLayer.h" +#include "paddle/utils/Logging.h" #ifndef PADDLE_ONLY_CPU #include "CudnnPoolLayer.h" #endif diff --git a/paddle/gserver/layers/PoolLayer.h b/paddle/gserver/layers/PoolLayer.h index c05d7a364d..318b89d7c2 100644 --- a/paddle/gserver/layers/PoolLayer.h +++ b/paddle/gserver/layers/PoolLayer.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once +#include #include "Layer.h" -#include "paddle/math/Matrix.h" #include "paddle/math/MathUtils.h" -#include +#include "paddle/math/Matrix.h" namespace paddle { diff --git a/paddle/gserver/layers/PoolProjectionLayer.cpp b/paddle/gserver/layers/PoolProjectionLayer.cpp index 392c548d45..ed5011ab89 100644 --- a/paddle/gserver/layers/PoolProjectionLayer.cpp +++ b/paddle/gserver/layers/PoolProjectionLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "PoolProjectionLayer.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "PoolProjectionLayer.h" namespace paddle { diff --git a/paddle/gserver/layers/PowerLayer.cpp b/paddle/gserver/layers/PowerLayer.cpp index eb69249270..64fecab5b0 100644 --- a/paddle/gserver/layers/PowerLayer.cpp +++ b/paddle/gserver/layers/PowerLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/RecurrentLayer.cpp b/paddle/gserver/layers/RecurrentLayer.cpp index 0832eeaa10..9f3bf76a2d 100644 --- a/paddle/gserver/layers/RecurrentLayer.cpp +++ b/paddle/gserver/layers/RecurrentLayer.cpp @@ -13,9 +13,9 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "Layer.h" -#include "paddle/utils/Stat.h" #include "SequenceToBatch.h" #include "paddle/utils/CommandLineParser.h" +#include "paddle/utils/Stat.h" P_DEFINE_bool(rnn_use_batch, false, "Using the batch method for calculation."); diff --git a/paddle/gserver/layers/RecurrentLayerGroup.cpp b/paddle/gserver/layers/RecurrentLayerGroup.cpp index 5cb4220623..af8dd61d84 100644 --- a/paddle/gserver/layers/RecurrentLayerGroup.cpp +++ b/paddle/gserver/layers/RecurrentLayerGroup.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/gserver/layers/Layer.h" #include +#include "paddle/gserver/layers/Layer.h" #include "paddle/gserver/gradientmachines/RecurrentGradientMachine.h" #include "paddle/utils/Stat.h" diff --git a/paddle/gserver/layers/ResizeLayer.cpp b/paddle/gserver/layers/ResizeLayer.cpp index e79732155a..7fcb3adea0 100644 --- a/paddle/gserver/layers/ResizeLayer.cpp +++ b/paddle/gserver/layers/ResizeLayer.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "Layer.h" -#include "paddle/math/Matrix.h" #include "paddle/math/BaseMatrix.h" +#include "paddle/math/Matrix.h" namespace paddle { /** diff --git a/paddle/gserver/layers/ScalingLayer.cpp b/paddle/gserver/layers/ScalingLayer.cpp index 013bff6b98..7f0084be6b 100644 --- a/paddle/gserver/layers/ScalingLayer.cpp +++ b/paddle/gserver/layers/ScalingLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp b/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp index 75d9fa8a97..9200a01eee 100644 --- a/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp +++ b/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "SelectiveFullyConnectedLayer.h" +#include +#include +#include "paddle/math/SparseMatrix.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" -#include "paddle/math/SparseMatrix.h" -#include -#include namespace paddle { diff --git a/paddle/gserver/layers/SequenceConcatLayer.cpp b/paddle/gserver/layers/SequenceConcatLayer.cpp index d3e0e16e96..069bc26e60 100644 --- a/paddle/gserver/layers/SequenceConcatLayer.cpp +++ b/paddle/gserver/layers/SequenceConcatLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/SequencePoolLayer.cpp b/paddle/gserver/layers/SequencePoolLayer.cpp index 856c889e3b..35260ca912 100644 --- a/paddle/gserver/layers/SequencePoolLayer.cpp +++ b/paddle/gserver/layers/SequencePoolLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "SequencePoolLayer.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/gserver/layers/SequenceReshapeLayer.cpp b/paddle/gserver/layers/SequenceReshapeLayer.cpp index 4b90424215..23924b0490 100644 --- a/paddle/gserver/layers/SequenceReshapeLayer.cpp +++ b/paddle/gserver/layers/SequenceReshapeLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/SequenceToBatch.cpp b/paddle/gserver/layers/SequenceToBatch.cpp index c12ed82197..5fa7b6f488 100644 --- a/paddle/gserver/layers/SequenceToBatch.cpp +++ b/paddle/gserver/layers/SequenceToBatch.cpp @@ -12,11 +12,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include -#include #include "SequenceToBatch.h" -#include #include +#include +#include +#include namespace paddle { diff --git a/paddle/gserver/layers/SequenceToBatch.h b/paddle/gserver/layers/SequenceToBatch.h index fe9b34b224..17e735a135 100644 --- a/paddle/gserver/layers/SequenceToBatch.h +++ b/paddle/gserver/layers/SequenceToBatch.h @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once -#include "paddle/math/Vector.h" #include "paddle/math/Matrix.h" +#include "paddle/math/Vector.h" namespace paddle { diff --git a/paddle/gserver/layers/SlopeInterceptLayer.cpp b/paddle/gserver/layers/SlopeInterceptLayer.cpp index 5c00e54f8c..b678f414b6 100644 --- a/paddle/gserver/layers/SlopeInterceptLayer.cpp +++ b/paddle/gserver/layers/SlopeInterceptLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/SubSequenceLayer.cpp b/paddle/gserver/layers/SubSequenceLayer.cpp index 8b35456391..c52fbee262 100644 --- a/paddle/gserver/layers/SubSequenceLayer.cpp +++ b/paddle/gserver/layers/SubSequenceLayer.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" #include "paddle/math/Vector.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/SumToOneNormLayer.cpp b/paddle/gserver/layers/SumToOneNormLayer.cpp index e6759171cb..aa99b49380 100644 --- a/paddle/gserver/layers/SumToOneNormLayer.cpp +++ b/paddle/gserver/layers/SumToOneNormLayer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Layer.h" #include "paddle/math/Matrix.h" +#include "paddle/utils/Logging.h" #include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/TransLayer.cpp b/paddle/gserver/layers/TransLayer.cpp index 5cbaaf8f08..d1fa90f384 100644 --- a/paddle/gserver/layers/TransLayer.cpp +++ b/paddle/gserver/layers/TransLayer.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "TransLayer.h" +#include "paddle/utils/Logging.h" namespace paddle { REGISTER_LAYER(trans, TransLayer); diff --git a/paddle/gserver/layers/TransLayer.h b/paddle/gserver/layers/TransLayer.h index 8189700759..b43fa1ebfb 100644 --- a/paddle/gserver/layers/TransLayer.h +++ b/paddle/gserver/layers/TransLayer.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once +#include #include "Layer.h" #include "paddle/math/Matrix.h" -#include namespace paddle { /** diff --git a/paddle/gserver/layers/TransposedFullMatrixProjection.cpp b/paddle/gserver/layers/TransposedFullMatrixProjection.cpp index 8282584ab4..3f7ff04882 100644 --- a/paddle/gserver/layers/TransposedFullMatrixProjection.cpp +++ b/paddle/gserver/layers/TransposedFullMatrixProjection.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Stat.h" #include "Projection.h" +#include "paddle/utils/Stat.h" namespace paddle { diff --git a/paddle/gserver/layers/ValidationLayer.cpp b/paddle/gserver/layers/ValidationLayer.cpp index f029ea4c51..5127bcaba3 100644 --- a/paddle/gserver/layers/ValidationLayer.cpp +++ b/paddle/gserver/layers/ValidationLayer.cpp @@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include #include +#include -#include "paddle/utils/Logging.h" #include "ValidationLayer.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/gserver/layers/ValidationLayer.h b/paddle/gserver/layers/ValidationLayer.h index f9c61503aa..471055429d 100644 --- a/paddle/gserver/layers/ValidationLayer.h +++ b/paddle/gserver/layers/ValidationLayer.h @@ -15,8 +15,8 @@ limitations under the License. */ #pragma once #include -#include "paddle/gserver/evaluators/Evaluator.h" #include "Layer.h" +#include "paddle/gserver/evaluators/Evaluator.h" P_DECLARE_int32(trainer_id); diff --git a/paddle/gserver/tests/LayerGradUtil.h b/paddle/gserver/tests/LayerGradUtil.h index 2b8f334f19..62ac2d160f 100644 --- a/paddle/gserver/tests/LayerGradUtil.h +++ b/paddle/gserver/tests/LayerGradUtil.h @@ -13,9 +13,9 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once -#include "paddle/trainer/Trainer.h" -#include "paddle/gserver/layers/DataLayer.h" #include "ModelConfig.pb.h" +#include "paddle/gserver/layers/DataLayer.h" +#include "paddle/trainer/Trainer.h" #include "TestUtil.h" using namespace std; // NOLINT diff --git a/paddle/gserver/tests/TestUtil.cpp b/paddle/gserver/tests/TestUtil.cpp index dc00711697..e656da5b8f 100644 --- a/paddle/gserver/tests/TestUtil.cpp +++ b/paddle/gserver/tests/TestUtil.cpp @@ -14,8 +14,8 @@ limitations under the License. */ #include "TestUtil.h" -#include "paddle/utils/CommandLineParser.h" #include "paddle/math/SparseMatrix.h" +#include "paddle/utils/CommandLineParser.h" P_DEFINE_int32(fixed_seq_length, 0, "Produce some sequence of fixed length"); @@ -63,8 +63,8 @@ MatrixPtr makeRandomSparseMatrix(size_t height, std::dynamic_pointer_cast(mat)->copyFrom( ids.data(), indices.data(), data.data(), HPPL_STREAM_DEFAULT); } else { - std::dynamic_pointer_cast(mat) - ->copyFrom(ids.data(), indices.data(), data.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + ids.data(), indices.data(), data.data()); } return mat; } else { @@ -80,8 +80,8 @@ MatrixPtr makeRandomSparseMatrix(size_t height, std::dynamic_pointer_cast(mat)->copyFrom( ids.data(), indices.data(), data.data(), HPPL_STREAM_DEFAULT); } else { - std::dynamic_pointer_cast(mat) - ->copyFrom(ids.data(), indices.data(), data.data()); + std::dynamic_pointer_cast(mat)->copyFrom( + ids.data(), indices.data(), data.data()); } return mat; } diff --git a/paddle/gserver/tests/test_ActivationGrad.cpp b/paddle/gserver/tests/test_ActivationGrad.cpp index 0181d62519..20a6126d0b 100644 --- a/paddle/gserver/tests/test_ActivationGrad.cpp +++ b/paddle/gserver/tests/test_ActivationGrad.cpp @@ -13,14 +13,14 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include #include -#include "paddle/gserver/layers/DataLayer.h" +#include #include "ModelConfig.pb.h" +#include "paddle/gserver/layers/DataLayer.h" #include "paddle/trainer/Trainer.h" -#include "TestUtil.h" #include "LayerGradUtil.h" +#include "TestUtil.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/gserver/tests/test_BatchNorm.cpp b/paddle/gserver/tests/test_BatchNorm.cpp index 8575999aba..3bd4e321b7 100644 --- a/paddle/gserver/tests/test_BatchNorm.cpp +++ b/paddle/gserver/tests/test_BatchNorm.cpp @@ -13,16 +13,16 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include #include -#include "paddle/gserver/layers/DataLayer.h" +#include #include "ModelConfig.pb.h" +#include "paddle/gserver/layers/DataLayer.h" +#include "paddle/gserver/layers/ExpandConvTransLayer.h" #include "paddle/trainer/Trainer.h" #include "paddle/utils/GlobalConstants.h" -#include "paddle/gserver/layers/ExpandConvTransLayer.h" -#include "TestUtil.h" #include "LayerGradUtil.h" +#include "TestUtil.h" using namespace paddle; // NOLINT using namespace std; // NOLINT @@ -35,80 +35,87 @@ P_DECLARE_bool(prev_batch_state); // Test that the batchNormLayer can be followed by a ConvLayer TEST(Layer, batchNorm) { - FLAGS_use_gpu = false; - TestConfig configBN; - const int CHANNELS = 6272; - const int IMG_SIZE = 1; - configBN.layerConfig.set_type("batch_norm"); - configBN.layerConfig.set_name("bn"); - configBN.layerConfig.set_size(CHANNELS * IMG_SIZE * IMG_SIZE); - configBN.layerConfig.set_active_type("relu"); - configBN.biasSize = CHANNELS; - configBN.inputDefs.push_back({INPUT_DATA, "layer_0", + FLAGS_use_gpu = false; + TestConfig configBN; + const int CHANNELS = 6272; + const int IMG_SIZE = 1; + configBN.layerConfig.set_type("batch_norm"); + configBN.layerConfig.set_name("bn"); + configBN.layerConfig.set_size(CHANNELS * IMG_SIZE * IMG_SIZE); + configBN.layerConfig.set_active_type("relu"); + configBN.biasSize = CHANNELS; + configBN.inputDefs.push_back({INPUT_DATA, + "layer_0", /* dim= */ IMG_SIZE * IMG_SIZE * CHANNELS, /* paraSize= */ CHANNELS}); - configBN.inputDefs.push_back({INPUT_DATA, "layer_1_running_mean", - 1, CHANNELS}); - configBN.inputDefs.back().isStatic = true; - configBN.inputDefs.push_back({INPUT_DATA, "layer_2_running_var", - 1, CHANNELS}); - configBN.inputDefs.back().isStatic = true; - - LayerInputConfig* input = configBN.layerConfig.add_inputs(); - configBN.layerConfig.add_inputs(); - configBN.layerConfig.add_inputs(); - - ImageConfig* img_conf = input->mutable_image_conf(); - img_conf->set_channels(CHANNELS); - img_conf->set_img_size(IMG_SIZE); - - // Setting up conv-layer config - TestConfig config; - config.biasSize = 64; - config.layerConfig.set_type("exconv"); - config.layerConfig.set_num_filters(64); - config.layerConfig.set_partial_sum(1); - config.layerConfig.set_shared_biases(true); - - config.inputDefs.push_back({INPUT_DATA, "bn", 6272, 204800}); - input = config.layerConfig.add_inputs(); - ConvConfig* conv = input->mutable_conv_conf(); - conv->set_filter_size(5); - conv->set_filter_size_y(5); - conv->set_channels(128); - conv->set_padding(1); - conv->set_padding_y(1); - conv->set_stride(2); - conv->set_stride_y(2); - conv->set_groups(1); - conv->set_filter_channels(conv->channels() / conv->groups()); - conv->set_img_size(7); - conv->set_output_x(3); - config.layerConfig.set_size(conv->output_x() * conv->output_x() * - config.layerConfig.num_filters()); - config.layerConfig.set_name("conv"); - - // data layer initialize - std::vector dataLayers; - LayerMap layerMap; - vector datas; - initDataLayer(configBN, &dataLayers, &datas, &layerMap, "batch_norm", - 100, false, false); - // test layer initialize - std::vector parameters; - LayerPtr bnLayer; - initTestLayer(configBN, &layerMap, ¶meters, &bnLayer); - - std::vector parameters2; - LayerPtr convLayer; - initTestLayer(config, &layerMap, ¶meters2, &convLayer); - - bnLayer->forward(PASS_GC); - convLayer->forward(PASS_GC); - - CHECK_EQ(convLayer->getOutputValue()->getHeight(), 100); - CHECK_EQ(convLayer->getOutputValue()->getWidth(), 576); + configBN.inputDefs.push_back( + {INPUT_DATA, "layer_1_running_mean", 1, CHANNELS}); + configBN.inputDefs.back().isStatic = true; + configBN.inputDefs.push_back( + {INPUT_DATA, "layer_2_running_var", 1, CHANNELS}); + configBN.inputDefs.back().isStatic = true; + + LayerInputConfig* input = configBN.layerConfig.add_inputs(); + configBN.layerConfig.add_inputs(); + configBN.layerConfig.add_inputs(); + + ImageConfig* img_conf = input->mutable_image_conf(); + img_conf->set_channels(CHANNELS); + img_conf->set_img_size(IMG_SIZE); + + // Setting up conv-layer config + TestConfig config; + config.biasSize = 64; + config.layerConfig.set_type("exconv"); + config.layerConfig.set_num_filters(64); + config.layerConfig.set_partial_sum(1); + config.layerConfig.set_shared_biases(true); + + config.inputDefs.push_back({INPUT_DATA, "bn", 6272, 204800}); + input = config.layerConfig.add_inputs(); + ConvConfig* conv = input->mutable_conv_conf(); + conv->set_filter_size(5); + conv->set_filter_size_y(5); + conv->set_channels(128); + conv->set_padding(1); + conv->set_padding_y(1); + conv->set_stride(2); + conv->set_stride_y(2); + conv->set_groups(1); + conv->set_filter_channels(conv->channels() / conv->groups()); + conv->set_img_size(7); + conv->set_output_x(3); + config.layerConfig.set_size(conv->output_x() * conv->output_x() * + config.layerConfig.num_filters()); + config.layerConfig.set_name("conv"); + + // data layer initialize + std::vector dataLayers; + LayerMap layerMap; + vector datas; + initDataLayer(configBN, + &dataLayers, + &datas, + &layerMap, + "batch_norm", + 100, + false, + false); + // test layer initialize + std::vector parameters; + LayerPtr bnLayer; + initTestLayer(configBN, &layerMap, ¶meters, &bnLayer); + + std::vector parameters2; + LayerPtr convLayer; + initTestLayer(config, &layerMap, ¶meters2, &convLayer); + + bnLayer->forward(PASS_GC); + convLayer->forward(PASS_GC); + + CHECK_EQ(convLayer->getOutputValue()->getHeight(), 100); + CHECK_EQ(convLayer->getOutputValue()->getWidth(), 576); } int main(int argc, char** argv) { diff --git a/paddle/gserver/tests/test_ConvTrans.cpp b/paddle/gserver/tests/test_ConvTrans.cpp index 3af3f08f40..83100e3bec 100644 --- a/paddle/gserver/tests/test_ConvTrans.cpp +++ b/paddle/gserver/tests/test_ConvTrans.cpp @@ -13,17 +13,17 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include #include -#include "paddle/gserver/layers/DataLayer.h" +#include #include "ModelConfig.pb.h" -#include "paddle/trainer/Trainer.h" -#include "paddle/utils/GlobalConstants.h" +#include "paddle/gserver/layers/DataLayer.h" #include "paddle/gserver/layers/ExpandConvTransLayer.h" #include "paddle/math/MathUtils.h" +#include "paddle/trainer/Trainer.h" +#include "paddle/utils/GlobalConstants.h" -#include "TestUtil.h" #include "LayerGradUtil.h" +#include "TestUtil.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/gserver/tests/test_ConvUnify.cpp b/paddle/gserver/tests/test_ConvUnify.cpp index d59acf96ac..02763406a3 100644 --- a/paddle/gserver/tests/test_ConvUnify.cpp +++ b/paddle/gserver/tests/test_ConvUnify.cpp @@ -13,17 +13,17 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include #include -#include "paddle/gserver/layers/DataLayer.h" +#include #include "ModelConfig.pb.h" -#include "paddle/trainer/Trainer.h" -#include "paddle/utils/GlobalConstants.h" +#include "paddle/gserver/layers/DataLayer.h" #include "paddle/gserver/layers/ExpandConvTransLayer.h" #include "paddle/math/MathUtils.h" +#include "paddle/trainer/Trainer.h" +#include "paddle/utils/GlobalConstants.h" -#include "TestUtil.h" #include "LayerGradUtil.h" +#include "TestUtil.h" using namespace paddle; // NOLINT using namespace std; // NOLINT @@ -36,10 +36,17 @@ P_DECLARE_bool(prev_batch_state); // Do one forward pass of convTrans layer and check to see if its output // matches the given result -MatrixPtr doOneConvTest(size_t imgSize, size_t output_x, size_t stride, - size_t padding, size_t filter_size, size_t channel, - size_t numfilters, size_t groups, MatrixPtr& inputData, - real* param, bool useGpu) { +MatrixPtr doOneConvTest(size_t imgSize, + size_t output_x, + size_t stride, + size_t padding, + size_t filter_size, + size_t channel, + size_t numfilters, + size_t groups, + MatrixPtr& inputData, + real* param, + bool useGpu) { TestConfig config; config.biasSize = numfilters; if (useGpu) { @@ -51,11 +58,10 @@ MatrixPtr doOneConvTest(size_t imgSize, size_t output_x, size_t stride, config.layerConfig.set_partial_sum(1); config.layerConfig.set_shared_biases(true); - size_t weightSize = channel* filter_size * filter_size * - config.layerConfig.num_filters() / groups; - config.inputDefs.push_back({INPUT_DATA, "layer_0", - imgSize * imgSize * channel, - weightSize}); + size_t weightSize = channel * filter_size * filter_size * + config.layerConfig.num_filters() / groups; + config.inputDefs.push_back( + {INPUT_DATA, "layer_0", imgSize * imgSize * channel, weightSize}); LayerInputConfig* input = config.layerConfig.add_inputs(); ConvConfig* conv = input->mutable_conv_conf(); conv->set_filter_size(filter_size); @@ -66,7 +72,7 @@ MatrixPtr doOneConvTest(size_t imgSize, size_t output_x, size_t stride, conv->set_stride(stride); conv->set_stride_y(stride); conv->set_groups(groups); - conv->set_filter_channels(channel/groups); + conv->set_filter_channels(channel / groups); conv->set_img_size(imgSize); conv->set_output_x(output_x); @@ -77,8 +83,8 @@ MatrixPtr doOneConvTest(size_t imgSize, size_t output_x, size_t stride, std::vector dataLayers; LayerMap layerMap; vector datas; - initDataLayer(config, &dataLayers, &datas, &layerMap, "conv", - 1, false, useGpu); + initDataLayer( + config, &dataLayers, &datas, &layerMap, "conv", 1, false, useGpu); dataLayers[0]->getOutputValue()->zeroMem(); dataLayers[0]->getOutputValue()->copyFrom(*inputData); @@ -88,106 +94,124 @@ MatrixPtr doOneConvTest(size_t imgSize, size_t output_x, size_t stride, initTestLayer(config, &layerMap, ¶meters, &convLayer); convLayer->getBiasParameter()->zeroMem(); convLayer->getParameters()[0]->zeroMem(); - convLayer->getParameters()[0]->getBuf(PARAMETER_VALUE)->copyFrom(param, - weightSize); + convLayer->getParameters()[0] + ->getBuf(PARAMETER_VALUE) + ->copyFrom(param, weightSize); convLayer->forward(PASS_GC); return convLayer->getOutputValue(); } TEST(Layer, convParaUnified) { - #ifndef PADDLE_ONLY_CPU - MatrixPtr input, resultCpu, resultGpu; - input = Matrix::create(1, 4 * 4, false, false); - float inputData[] = {1, 2, 3, 4, - 5, 6, 7, 8, - 9, 10, 11, 12, - 13, 14, 15, 16}; - float param[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, - 9, 8, 7, 6, 5, 4, 3, 2, 1}; - - input->setData(inputData); - - resultCpu = doOneConvTest(/* imgSize */ 4, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 3, - /*channel*/ 1, - /*numfilters*/ 2, - /*groups*/ 1, - input, param, false); - - resultGpu = doOneConvTest(/* imgSize */ 4, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 3, - /*channel*/ 1, - /*numfilters*/ 2, - /*groups*/ 1, - input, param, true); - checkMatrixEqual(resultCpu, resultGpu); - - input = Matrix::create(1, 3 * 3 * 2, false, false); - float inputData2[] = {1, 2, 3, - 4, 5, 6, - 7, 8, 9, - - 10, 11, 12, - 13, 14, 15, - 16, 17, 18}; - float param2[] = {1, 2, 3, 4, 5, 6, 7, 8, - 8, 7, 6, 5, 4, 3, 2, 1}; - - input->setData(inputData2); - - resultCpu = doOneConvTest(/* imgSize */ 3, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 2, - /*channel*/ 2, - /*numfilters*/ 2, - /*groups*/ 1, - input, param2, false); - - resultGpu = doOneConvTest(/* imgSize */ 3, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 2, - /*channel*/ 2, - /*numfilters*/ 2, - /*groups*/ 1, - input, param2, true); - checkMatrixEqual(resultCpu, resultGpu); - - - float param3[] = {1, 2, 3, 4, - 4, 3, 2, 1}; - - resultCpu = doOneConvTest(/* imgSize */ 3, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 2, - /*channel*/ 2, - /*numfilters*/ 2, - /*groups*/ 2, - input, param3, false); - - resultGpu = doOneConvTest(/* imgSize */ 3, - /* output_x */ 2, - /* stride */ 1, - /* padding */ 0, - /* filter_size */ 2, - /*channel*/ 2, - /*numfilters*/ 2, - /*groups*/ 2, - input, param3, true); - checkMatrixEqual(resultCpu, resultGpu); - #endif +#ifndef PADDLE_ONLY_CPU + MatrixPtr input, resultCpu, resultGpu; + input = Matrix::create(1, 4 * 4, false, false); + float inputData[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}; + float param[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 9, 8, 7, 6, 5, 4, 3, 2, 1}; + + input->setData(inputData); + + resultCpu = doOneConvTest(/* imgSize */ 4, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 3, + /*channel*/ 1, + /*numfilters*/ 2, + /*groups*/ 1, + input, + param, + false); + + resultGpu = doOneConvTest(/* imgSize */ 4, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 3, + /*channel*/ 1, + /*numfilters*/ 2, + /*groups*/ 1, + input, + param, + true); + checkMatrixEqual(resultCpu, resultGpu); + + input = Matrix::create(1, 3 * 3 * 2, false, false); + float inputData2[] = {1, + 2, + 3, + 4, + 5, + 6, + 7, + 8, + 9, + + 10, + 11, + 12, + 13, + 14, + 15, + 16, + 17, + 18}; + float param2[] = {1, 2, 3, 4, 5, 6, 7, 8, 8, 7, 6, 5, 4, 3, 2, 1}; + + input->setData(inputData2); + + resultCpu = doOneConvTest(/* imgSize */ 3, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 2, + /*channel*/ 2, + /*numfilters*/ 2, + /*groups*/ 1, + input, + param2, + false); + + resultGpu = doOneConvTest(/* imgSize */ 3, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 2, + /*channel*/ 2, + /*numfilters*/ 2, + /*groups*/ 1, + input, + param2, + true); + checkMatrixEqual(resultCpu, resultGpu); + + float param3[] = {1, 2, 3, 4, 4, 3, 2, 1}; + + resultCpu = doOneConvTest(/* imgSize */ 3, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 2, + /*channel*/ 2, + /*numfilters*/ 2, + /*groups*/ 2, + input, + param3, + false); + + resultGpu = doOneConvTest(/* imgSize */ 3, + /* output_x */ 2, + /* stride */ 1, + /* padding */ 0, + /* filter_size */ 2, + /*channel*/ 2, + /*numfilters*/ 2, + /*groups*/ 2, + input, + param3, + true); + checkMatrixEqual(resultCpu, resultGpu); +#endif } int main(int argc, char** argv) { diff --git a/paddle/gserver/tests/test_Evaluator.cpp b/paddle/gserver/tests/test_Evaluator.cpp index 2c20f3a52f..7a930aebcf 100644 --- a/paddle/gserver/tests/test_Evaluator.cpp +++ b/paddle/gserver/tests/test_Evaluator.cpp @@ -15,8 +15,8 @@ limitations under the License. */ #include #include #include "ModelConfig.pb.h" -#include "paddle/trainer/Trainer.h" #include "TestUtil.h" +#include "paddle/trainer/Trainer.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/gserver/tests/test_LayerGrad.cpp b/paddle/gserver/tests/test_LayerGrad.cpp index 7983d9fe64..9f8b197df5 100644 --- a/paddle/gserver/tests/test_LayerGrad.cpp +++ b/paddle/gserver/tests/test_LayerGrad.cpp @@ -17,8 +17,8 @@ limitations under the License. */ #include #include "ModelConfig.pb.h" #include "paddle/gserver/layers/DataLayer.h" -#include "paddle/trainer/Trainer.h" #include "paddle/math/MathUtils.h" +#include "paddle/trainer/Trainer.h" #include "LayerGradUtil.h" #include "TestUtil.h" diff --git a/paddle/gserver/tests/test_MultinomialSampler.cpp b/paddle/gserver/tests/test_MultinomialSampler.cpp index fc164da8ea..eadf40ade0 100644 --- a/paddle/gserver/tests/test_MultinomialSampler.cpp +++ b/paddle/gserver/tests/test_MultinomialSampler.cpp @@ -20,8 +20,8 @@ limitations under the License. */ #undef PADDLE_DISABLE_TIMER #include "paddle/utils/Stat.h" -#include "paddle/utils/Util.h" #include "paddle/gserver/layers/MultinomialSampler.h" +#include "paddle/utils/Util.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/gserver/tests/test_NetworkCompare.cpp b/paddle/gserver/tests/test_NetworkCompare.cpp index ff6b5ab0d0..baa55aa025 100644 --- a/paddle/gserver/tests/test_NetworkCompare.cpp +++ b/paddle/gserver/tests/test_NetworkCompare.cpp @@ -13,14 +13,14 @@ See the License for the specific language governing permissions and limitations under the License. */ #undef PADDLE_DISABLE_TIMER +#include #include -#include #include -#include +#include +#include "TestUtil.h" #include "paddle/trainer/Trainer.h" #include "paddle/utils/Stat.h" -#include "TestUtil.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/gserver/tests/test_ProtoDataProvider.cpp b/paddle/gserver/tests/test_ProtoDataProvider.cpp index d5b8017cd1..d421b6e2f2 100644 --- a/paddle/gserver/tests/test_ProtoDataProvider.cpp +++ b/paddle/gserver/tests/test_ProtoDataProvider.cpp @@ -17,8 +17,8 @@ limitations under the License. */ #include -#include "paddle/utils/Util.h" #include "paddle/gserver/dataproviders/ProtoDataProvider.h" +#include "paddle/utils/Util.h" #include "TestUtil.h" diff --git a/paddle/gserver/tests/test_RecurrentLayer.cpp b/paddle/gserver/tests/test_RecurrentLayer.cpp index 3f26b710e9..cd96ca7c84 100644 --- a/paddle/gserver/tests/test_RecurrentLayer.cpp +++ b/paddle/gserver/tests/test_RecurrentLayer.cpp @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include #include +#include +#include "ModelConfig.pb.h" #include "paddle/gserver/layers/DataLayer.h" #include "paddle/gserver/layers/Layer.h" -#include "ModelConfig.pb.h" #include "TestUtil.h" @@ -220,8 +220,8 @@ TEST(Layer, RecurrentLayer) { } #define protected public -#include "paddle/gserver/layers/LstmLayer.h" #include "paddle/gserver/layers/GatedRecurrentLayer.h" +#include "paddle/gserver/layers/LstmLayer.h" template class TestRecurrentLayer { public: diff --git a/paddle/gserver/tests/test_SelectiveFCLayer.cpp b/paddle/gserver/tests/test_SelectiveFCLayer.cpp index c588f69446..4f3a95a535 100644 --- a/paddle/gserver/tests/test_SelectiveFCLayer.cpp +++ b/paddle/gserver/tests/test_SelectiveFCLayer.cpp @@ -12,17 +12,17 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include +#include #include +#include #include #include -#include -#include -#include +#include "ModelConfig.pb.h" #include "paddle/gserver/layers/DataLayer.h" -#include "paddle/gserver/layers/Layer.h" #include "paddle/gserver/layers/FullyConnectedLayer.h" +#include "paddle/gserver/layers/Layer.h" #include "paddle/gserver/layers/SelectiveFullyConnectedLayer.h" -#include "ModelConfig.pb.h" #include "paddle/math/CpuSparseMatrix.h" #include "paddle/trainer/Trainer.h" diff --git a/paddle/gserver/tests/test_WarpCTCLayer.cpp b/paddle/gserver/tests/test_WarpCTCLayer.cpp index e526a27906..700425412c 100644 --- a/paddle/gserver/tests/test_WarpCTCLayer.cpp +++ b/paddle/gserver/tests/test_WarpCTCLayer.cpp @@ -14,11 +14,11 @@ limitations under the License. */ #include #include -#include "paddle/gserver/layers/Layer.h" -#include "paddle/gserver/layers/DataLayer.h" +#include "ModelConfig.pb.h" #include "paddle/gserver/layers/CTCLayer.h" +#include "paddle/gserver/layers/DataLayer.h" +#include "paddle/gserver/layers/Layer.h" #include "paddle/gserver/layers/WarpCTCLayer.h" -#include "ModelConfig.pb.h" #include "TestUtil.h" diff --git a/paddle/math/Allocator.h b/paddle/math/Allocator.h index 4d0a1506be..666a8b8368 100644 --- a/paddle/math/Allocator.h +++ b/paddle/math/Allocator.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include #include +#include #include "hl_gpu.h" #include "paddle/utils/Logging.h" diff --git a/paddle/math/BaseMatrix.h b/paddle/math/BaseMatrix.h index 368557bb26..2933c20fba 100644 --- a/paddle/math/BaseMatrix.h +++ b/paddle/math/BaseMatrix.h @@ -13,10 +13,10 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once -#include #include -#include "paddle/utils/TypeDefs.h" +#include #include "TensorExpression.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/math/CpuSparseMatrix.cpp b/paddle/math/CpuSparseMatrix.cpp index 324c7ec0ca..b5d5b6ef61 100644 --- a/paddle/math/CpuSparseMatrix.cpp +++ b/paddle/math/CpuSparseMatrix.cpp @@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "hl_gpu.h" #include "CpuSparseMatrix.h" #include "SparseMatrix.h" +#include "float.h" +#include "hl_gpu.h" #include "paddle/math/MathUtils.h" #include "paddle/utils/Util.h" -#include "float.h" namespace paddle { @@ -656,9 +656,9 @@ void CpuSparseMatrix::trimFrom(const CpuSparseMatrix& src) { if (format_ == SPARSE_CSR) { int* srcCols = src.getCols(); size_t numLessWidth = - std::count_if(srcCols, - srcCols + src.getElementCnt(), - [this](size_t n) { return n < this->width_; }); + std::count_if(srcCols, srcCols + src.getElementCnt(), [this](size_t n) { + return n < this->width_; + }); resize(height_, width_, numLessWidth, valueType_, format_); rows_[0] = 0; size_t index = 0; diff --git a/paddle/math/MathFunctions.cpp b/paddle/math/MathFunctions.cpp index 037525b402..d7aa118487 100644 --- a/paddle/math/MathFunctions.cpp +++ b/paddle/math/MathFunctions.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "MathFunctions.h" -#include "hl_matrix_ops.cuh" #include "hl_matrix_apply.cuh" +#include "hl_matrix_ops.cuh" namespace paddle { diff --git a/paddle/math/MathUtils.cpp b/paddle/math/MathUtils.cpp index 1fb7655c5a..5bbc3e4e37 100644 --- a/paddle/math/MathUtils.cpp +++ b/paddle/math/MathUtils.cpp @@ -14,8 +14,8 @@ limitations under the License. */ #include "MathUtils.h" #include -#include "paddle/utils/Logging.h" #include "Vector.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/math/Matrix.h b/paddle/math/Matrix.h index 395143a4b1..4342ca52a3 100644 --- a/paddle/math/Matrix.h +++ b/paddle/math/Matrix.h @@ -14,20 +14,20 @@ limitations under the License. */ #pragma once +#include #include #include -#include #include "paddle/utils/Logging.h" #include "paddle/utils/ThreadLocal.h" #include +#include "BaseMatrix.h" #include "MemoryHandle.h" -#include "paddle/utils/TypeDefs.h" #include "Vector.h" #include "paddle/utils/ThreadLocal.h" -#include "BaseMatrix.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/math/MatrixBitCode.cpp b/paddle/math/MatrixBitCode.cpp index 6390d4b6a5..cea912d3ca 100644 --- a/paddle/math/MatrixBitCode.cpp +++ b/paddle/math/MatrixBitCode.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" -#include "paddle/utils/Util.h" #include "Matrix.h" #include "hl_gpu.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/math/MemoryHandle.cpp b/paddle/math/MemoryHandle.cpp index 4c4a827b23..84afb5944c 100644 --- a/paddle/math/MemoryHandle.cpp +++ b/paddle/math/MemoryHandle.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include "MemoryHandle.h" +#include #include "Storage.h" namespace paddle { diff --git a/paddle/math/PoolAllocator.h b/paddle/math/PoolAllocator.h index 1544cb2cfc..c06efa9ac7 100644 --- a/paddle/math/PoolAllocator.h +++ b/paddle/math/PoolAllocator.h @@ -14,11 +14,11 @@ limitations under the License. */ #pragma once +#include #include #include -#include #include -#include +#include #include "Allocator.h" namespace paddle { diff --git a/paddle/math/SparseMatrix.cpp b/paddle/math/SparseMatrix.cpp index d2779cc9f5..9154503c21 100644 --- a/paddle/math/SparseMatrix.cpp +++ b/paddle/math/SparseMatrix.cpp @@ -12,13 +12,13 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "SparseMatrix.h" #include +#include #include #include "hl_gpu.h" -#include "SparseMatrix.h" -#include "paddle/utils/Util.h" #include "hl_top_k.h" -#include +#include "paddle/utils/Util.h" namespace paddle { @@ -537,11 +537,9 @@ void GpuSparseMatrix::transpose(MatrixPtr matTrans, bool memAlloc) { dataVec.emplace_back( rows.getData()[i], cols_full.getData()[i], value.getData()[i]); } - std::sort(dataVec.begin(), - dataVec.end(), - [](Element a, Element b) { - return a.row < b.row || (a.row == b.row && a.col < b.col); - }); + std::sort(dataVec.begin(), dataVec.end(), [](Element a, Element b) { + return a.row < b.row || (a.row == b.row && a.col < b.col); + }); /*get sorted data, row index, and col index, put them in the right place*/ cols.resize(height_ + 1); diff --git a/paddle/math/SparseMatrix.h b/paddle/math/SparseMatrix.h index f8d9ffc29f..bd96a3301d 100644 --- a/paddle/math/SparseMatrix.h +++ b/paddle/math/SparseMatrix.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once #include -#include "Matrix.h" #include "CpuSparseMatrix.h" +#include "Matrix.h" namespace paddle { diff --git a/paddle/math/SparseRowMatrix.h b/paddle/math/SparseRowMatrix.h index 2fee1b39fe..badb4b9c1c 100644 --- a/paddle/math/SparseRowMatrix.h +++ b/paddle/math/SparseRowMatrix.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once -#include #include -#include "paddle/utils/CommandLineParser.h" +#include #include "Matrix.h" +#include "paddle/utils/CommandLineParser.h" #include "paddle/utils/Util.h" P_DECLARE_bool(allow_inefficient_sparse_update); diff --git a/paddle/math/Storage.cpp b/paddle/math/Storage.cpp index 0170b4efb8..f9a2c12cd5 100644 --- a/paddle/math/Storage.cpp +++ b/paddle/math/Storage.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" -#include "Allocator.h" #include "Storage.h" +#include "Allocator.h" +#include "paddle/utils/Util.h" P_DEFINE_int32(pool_limit_size, 536870912, diff --git a/paddle/math/Storage.h b/paddle/math/Storage.h index 3658320182..06a66b5f14 100644 --- a/paddle/math/Storage.h +++ b/paddle/math/Storage.h @@ -16,8 +16,8 @@ limitations under the License. */ #include #include -#include "paddle/utils/Locks.h" #include "PoolAllocator.h" +#include "paddle/utils/Locks.h" namespace paddle { diff --git a/paddle/math/TensorEvaluate.h b/paddle/math/TensorEvaluate.h index 346ed7ab13..9de2099b85 100644 --- a/paddle/math/TensorEvaluate.h +++ b/paddle/math/TensorEvaluate.h @@ -15,8 +15,8 @@ limitations under the License. */ #pragma once #include -#include "paddle/utils/Logging.h" #include "hl_base.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/math/TensorExpression.h b/paddle/math/TensorExpression.h index 7f28ad83bb..9bd789e8c5 100644 --- a/paddle/math/TensorExpression.h +++ b/paddle/math/TensorExpression.h @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once -#include #include -#include "paddle/utils/TypeDefs.h" -#include "paddle/utils/Logging.h" +#include #include "hl_tensor_ops.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/math/TrainingAlgorithmOp.h b/paddle/math/TrainingAlgorithmOp.h index 2dc56f69e5..881a8d72d8 100644 --- a/paddle/math/TrainingAlgorithmOp.h +++ b/paddle/math/TrainingAlgorithmOp.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include "paddle/utils/Logging.h" #include "BaseMatrix.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/math/Vector.cpp b/paddle/math/Vector.cpp index 484f4c9252..eaa1cdce30 100644 --- a/paddle/math/Vector.cpp +++ b/paddle/math/Vector.cpp @@ -12,17 +12,17 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" #include "Vector.h" +#include "paddle/utils/Util.h" #include -#include "paddle/utils/Logging.h" -#include "paddle/utils/ThreadLocal.h" -#include "paddle/utils/Thread.h" -#include "paddle/utils/Flags.h" #include "Matrix.h" #include "hl_gpu.h" #include "hl_table_apply.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/Thread.h" +#include "paddle/utils/ThreadLocal.h" namespace paddle { @@ -754,8 +754,7 @@ void ParallelCpuVectorT::exec(SyncThreadPool::JobFunc func) { } template -CpuGpuVectorT::CpuGpuVectorT(size_t size, bool useGpu) - : sync_(nullptr) { +CpuGpuVectorT::CpuGpuVectorT(size_t size, bool useGpu) : sync_(nullptr) { if (!useGpu) { cpuVectorT_ = std::make_shared>(size); } else { diff --git a/paddle/math/Vector.h b/paddle/math/Vector.h index 535580ac37..8a24103bd4 100644 --- a/paddle/math/Vector.h +++ b/paddle/math/Vector.h @@ -14,15 +14,15 @@ limitations under the License. */ #pragma once -#include #include +#include #include -#include "MemoryHandle.h" -#include "paddle/utils/TypeDefs.h" #include "BaseMatrix.h" +#include "MemoryHandle.h" #include "paddle/utils/Thread.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/math/tests/OriginalOptimizerApi.h b/paddle/math/tests/OriginalOptimizerApi.h index ddcdd6bb51..0188372771 100644 --- a/paddle/math/tests/OriginalOptimizerApi.h +++ b/paddle/math/tests/OriginalOptimizerApi.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include "paddle/utils/GlobalConstants.h" #include "paddle/math/Vector.h" +#include "paddle/utils/GlobalConstants.h" using namespace paddle; // NOLINT diff --git a/paddle/math/tests/TestUtils.h b/paddle/math/tests/TestUtils.h index 5f9fab7245..c302096188 100644 --- a/paddle/math/tests/TestUtils.h +++ b/paddle/math/tests/TestUtils.h @@ -40,9 +40,9 @@ limitations under the License. */ */ #include +#include "TensorCheck.h" #include "paddle/math/Matrix.h" #include "paddle/math/SparseMatrix.h" -#include "TensorCheck.h" namespace autotest { diff --git a/paddle/math/tests/test_Allocator.cpp b/paddle/math/tests/test_Allocator.cpp index 440fcda0fe..33e0952efe 100644 --- a/paddle/math/tests/test_Allocator.cpp +++ b/paddle/math/tests/test_Allocator.cpp @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include "paddle/utils/Util.h" #include "paddle/utils/Logging.h" +#include "paddle/utils/Util.h" #define private public -#include "paddle/math/MemoryHandle.h" #include "paddle/math/Allocator.h" +#include "paddle/math/MemoryHandle.h" #include "paddle/math/PoolAllocator.h" using namespace paddle; // NOLINT diff --git a/paddle/math/tests/test_BaseMatrix.cpp b/paddle/math/tests/test_BaseMatrix.cpp index a4683918ca..cc7c1e7eb2 100644 --- a/paddle/math/tests/test_BaseMatrix.cpp +++ b/paddle/math/tests/test_BaseMatrix.cpp @@ -20,8 +20,8 @@ limitations under the License. */ */ #include -#include "paddle/math/BaseMatrix.h" #include "TestUtils.h" +#include "paddle/math/BaseMatrix.h" using paddle::BaseMatrix; using paddle::Matrix; diff --git a/paddle/math/tests/test_CpuGpuVector.cpp b/paddle/math/tests/test_CpuGpuVector.cpp index c671735875..624fa20ca5 100644 --- a/paddle/math/tests/test_CpuGpuVector.cpp +++ b/paddle/math/tests/test_CpuGpuVector.cpp @@ -14,10 +14,10 @@ limitations under the License. */ #ifndef PADDLE_ONLY_CPU -#include "paddle/utils/Util.h" +#include #include "paddle/math/Vector.h" +#include "paddle/utils/Util.h" #include "test_matrixUtil.h" -#include using namespace paddle; // NOLINT diff --git a/paddle/math/tests/test_ExecViaCpu.cpp b/paddle/math/tests/test_ExecViaCpu.cpp index b328ebf554..27216ddb58 100644 --- a/paddle/math/tests/test_ExecViaCpu.cpp +++ b/paddle/math/tests/test_ExecViaCpu.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include -#include +#include #include +#include #include "paddle/math/SparseMatrix.h" using namespace paddle; // NOLINT diff --git a/paddle/math/tests/test_GpuProfiler.cpp b/paddle/math/tests/test_GpuProfiler.cpp index e5fd6f4523..d490078d90 100644 --- a/paddle/math/tests/test_GpuProfiler.cpp +++ b/paddle/math/tests/test_GpuProfiler.cpp @@ -14,12 +14,12 @@ limitations under the License. */ #ifndef PADDLE_ONLY_CPU -#include "paddle/utils/Util.h" -#include "paddle/math/Matrix.h" -#include "paddle/math/SparseMatrix.h" #include #include "paddle/gserver/tests/TestUtil.h" +#include "paddle/math/Matrix.h" +#include "paddle/math/SparseMatrix.h" #include "paddle/utils/Stat.h" +#include "paddle/utils/Util.h" using namespace paddle; // NOLINT using namespace std; // NOLINT @@ -52,7 +52,9 @@ void MatrixCheckErr(const Matrix& matrix1, const Matrix& matrix2) { EXPECT_EQ(count, 0) << "There are " << count << " different element."; } -void testBilinearFwdBwd(int numSamples, int imgSizeH, int imgSizeW, +void testBilinearFwdBwd(int numSamples, + int imgSizeH, + int imgSizeW, int channels) { int inWidth = imgSizeH * imgSizeW * channels; int outWidth = 2 * imgSizeH * 2 * imgSizeW * channels; @@ -73,10 +75,22 @@ void testBilinearFwdBwd(int numSamples, int imgSizeH, int imgSizeW, { // nvprof: GPU Proflier REGISTER_GPU_PROFILER("testBilinearFwdBwd"); - target->bilinearForward(*input, imgSizeH, imgSizeW, - 2 * imgSizeH, 2 * imgSizeW, channels, ratioH, ratioW); - targetGpu->bilinearForward(*inputGpu, imgSizeH, imgSizeW, - 2 * imgSizeH, 2 * imgSizeW, channels, ratioH, ratioW); + target->bilinearForward(*input, + imgSizeH, + imgSizeW, + 2 * imgSizeH, + 2 * imgSizeW, + channels, + ratioH, + ratioW); + targetGpu->bilinearForward(*inputGpu, + imgSizeH, + imgSizeW, + 2 * imgSizeH, + 2 * imgSizeW, + channels, + ratioH, + ratioW); } // check @@ -88,8 +102,8 @@ void testBilinearFwdBwd(int numSamples, int imgSizeH, int imgSizeW, MatrixPtr inputGpuGrad = GpuMatrix::create(numSamples, inWidth, false, true); MatrixPtr targetGrad = CpuMatrix::create(numSamples, outWidth, false, false); - MatrixPtr targetGpuGrad = GpuMatrix::create(numSamples, outWidth, false, - true); + MatrixPtr targetGpuGrad = + GpuMatrix::create(numSamples, outWidth, false, true); MatrixPtr targetCheckGrad = CpuMatrix::create(numSamples, inWidth, false, false); @@ -98,10 +112,22 @@ void testBilinearFwdBwd(int numSamples, int imgSizeH, int imgSizeW, inputGpuGrad->copyFrom(*inputGrad); targetGpuGrad->copyFrom(*targetGrad); - inputGrad->bilinearBackward(*targetGrad, 2 * imgSizeH, 2 * imgSizeW, - imgSizeH, imgSizeW, channels, ratioH, ratioW); - inputGpuGrad->bilinearBackward(*targetGpuGrad, 2 * imgSizeH, 2 * imgSizeW, - imgSizeH, imgSizeW, channels, ratioH, ratioW); + inputGrad->bilinearBackward(*targetGrad, + 2 * imgSizeH, + 2 * imgSizeW, + imgSizeH, + imgSizeW, + channels, + ratioH, + ratioW); + inputGpuGrad->bilinearBackward(*targetGpuGrad, + 2 * imgSizeH, + 2 * imgSizeW, + imgSizeH, + imgSizeW, + channels, + ratioH, + ratioW); // check targetCheckGrad->copyFrom(*inputGpuGrad); @@ -116,8 +142,9 @@ TEST(Profiler, testBilinearFwdBwd) { // nvprof: GPU Proflier REGISTER_GPU_PROFILER("testBilinearFwdBwd"); // Paddle built-in timer - REGISTER_TIMER_INFO("testBilinearFwdBwd", - "numSamples = 10, channels = 16, imgSizeX = 64, imgSizeY = 64"); + REGISTER_TIMER_INFO( + "testBilinearFwdBwd", + "numSamples = 10, channels = 16, imgSizeX = 64, imgSizeY = 64"); testBilinearFwdBwd(numSamples, imgSize, imgSize, channels); } globalStat.printAllStatus(); @@ -128,8 +155,9 @@ int main(int argc, char** argv) { initMain(argc, argv); // nvprof: GPU Proflier - REGISTER_GPU_PROFILER("RecursiveProfilingTest", - "numSamples = 10, channels = 16, imgSizeX = 64, imgSizeY = 64"); + REGISTER_GPU_PROFILER( + "RecursiveProfilingTest", + "numSamples = 10, channels = 16, imgSizeX = 64, imgSizeY = 64"); return RUN_ALL_TESTS(); } diff --git a/paddle/math/tests/test_SIMDFunctions.cpp b/paddle/math/tests/test_SIMDFunctions.cpp index 2c54121d99..f62843310d 100644 --- a/paddle/math/tests/test_SIMDFunctions.cpp +++ b/paddle/math/tests/test_SIMDFunctions.cpp @@ -17,10 +17,10 @@ limitations under the License. */ #include -#include -#include #include +#include #include +#include #include #include diff --git a/paddle/math/tests/test_TrainingAlgorithm.cpp b/paddle/math/tests/test_TrainingAlgorithm.cpp index 93a930cc2f..1bf6a0cc43 100644 --- a/paddle/math/tests/test_TrainingAlgorithm.cpp +++ b/paddle/math/tests/test_TrainingAlgorithm.cpp @@ -13,11 +13,11 @@ See the License for the specific language governing permissions and limitations under the License. */ #include -#include "paddle/utils/Util.h" -#include "paddle/math/TrainingAlgorithmOp.h" #include "OriginalOptimizerApi.h" -#include "TensorCheck.h" #include "PerfUtils.h" +#include "TensorCheck.h" +#include "paddle/math/TrainingAlgorithmOp.h" +#include "paddle/utils/Util.h" using namespace paddle; // NOLINT diff --git a/paddle/math/tests/test_batchTranspose.cpp b/paddle/math/tests/test_batchTranspose.cpp index 88631c62b8..9925e24dc1 100644 --- a/paddle/math/tests/test_batchTranspose.cpp +++ b/paddle/math/tests/test_batchTranspose.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "test_matrixUtil.h" #include "hl_batch_transpose.h" +#include "test_matrixUtil.h" using namespace paddle; // NOLINT diff --git a/paddle/math/tests/test_matrixCompare.cpp b/paddle/math/tests/test_matrixCompare.cpp index 713792d82b..62de5b25e4 100644 --- a/paddle/math/tests/test_matrixCompare.cpp +++ b/paddle/math/tests/test_matrixCompare.cpp @@ -16,13 +16,13 @@ limitations under the License. */ /// This unittest checks GpuMatrix/CpuMatrix get same result, so disable when /// only cpu version. -#include "paddle/utils/Util.h" -#include "paddle/math/Matrix.h" -#include "paddle/math/SparseMatrix.h" #include +#include "TensorCheck.h" #include "paddle/gserver/tests/TestUtil.h" +#include "paddle/math/Matrix.h" +#include "paddle/math/SparseMatrix.h" #include "paddle/utils/Stat.h" -#include "TensorCheck.h" +#include "paddle/utils/Util.h" using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/math/tests/test_perturbation.cpp b/paddle/math/tests/test_perturbation.cpp index eaf4dfea66..60ebae0153 100644 --- a/paddle/math/tests/test_perturbation.cpp +++ b/paddle/math/tests/test_perturbation.cpp @@ -14,10 +14,10 @@ limitations under the License. */ #ifndef PADDLE_ONLY_CPU -#include +#include #include +#include #include -#include #include "hl_cuda.h" #include "hl_perturbation_util.cuh" diff --git a/paddle/math/tests/test_sparseMatrixCompare.cpp b/paddle/math/tests/test_sparseMatrixCompare.cpp index eff2c502bb..6f6de238ba 100644 --- a/paddle/math/tests/test_sparseMatrixCompare.cpp +++ b/paddle/math/tests/test_sparseMatrixCompare.cpp @@ -17,10 +17,10 @@ limitations under the License. */ // so disable when /// only cpu version. -#include "paddle/utils/Util.h" +#include #include "paddle/math/Matrix.h" +#include "paddle/utils/Util.h" #include "test_matrixUtil.h" -#include using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/parameter/Argument.cpp b/paddle/parameter/Argument.cpp index b632a11bbd..e91daa3717 100644 --- a/paddle/parameter/Argument.cpp +++ b/paddle/parameter/Argument.cpp @@ -551,11 +551,10 @@ void Argument::getSeqInfo(std::vector* seqInfo) const { } seqInfo->push_back(info); } - std::sort(seqInfo->begin(), - seqInfo->end(), - [](const SeqInfo& a, const SeqInfo& b) { - return a.topLevelLength > b.topLevelLength; - }); + std::sort( + seqInfo->begin(), seqInfo->end(), [](const SeqInfo& a, const SeqInfo& b) { + return a.topLevelLength > b.topLevelLength; + }); } void Argument::checkSubset() const { diff --git a/paddle/parameter/Argument.h b/paddle/parameter/Argument.h index 69d57a28c0..afd2de0202 100644 --- a/paddle/parameter/Argument.h +++ b/paddle/parameter/Argument.h @@ -18,9 +18,9 @@ limitations under the License. */ #include "paddle/math/Matrix.h" #include "paddle/math/Vector.h" +#include "paddle/parameter/Parameter.h" #include "paddle/utils/Locks.h" #include "paddle/utils/Util.h" -#include "paddle/parameter/Parameter.h" namespace paddle { diff --git a/paddle/parameter/FirstOrderOptimizer.cpp b/paddle/parameter/FirstOrderOptimizer.cpp index 17268d3715..630f15c8cf 100644 --- a/paddle/parameter/FirstOrderOptimizer.cpp +++ b/paddle/parameter/FirstOrderOptimizer.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" -#include "paddle/utils/Flags.h" -#include "paddle/math/TrainingAlgorithmOp.h" #include "FirstOrderOptimizer.h" +#include "paddle/math/TrainingAlgorithmOp.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Util.h" #include diff --git a/paddle/parameter/ParallelParameter.cpp b/paddle/parameter/ParallelParameter.cpp index b3182306a4..cea77e5b17 100644 --- a/paddle/parameter/ParallelParameter.cpp +++ b/paddle/parameter/ParallelParameter.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include +#include "paddle/utils/Logging.h" #include "ParallelParameter.h" diff --git a/paddle/parameter/ParallelParameter.h b/paddle/parameter/ParallelParameter.h index b0fe82d3c4..417e386dc7 100644 --- a/paddle/parameter/ParallelParameter.h +++ b/paddle/parameter/ParallelParameter.h @@ -16,19 +16,19 @@ limitations under the License. */ #include +#include +#include #include #include #include -#include -#include #include "hl_gpu.h" -#include "paddle/utils/Flags.h" -#include "paddle/utils/Locks.h" +#include "paddle/math/Vector.h" #include "paddle/parameter/Parameter.h" #include "paddle/parameter/ParameterUpdateFunctions.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Locks.h" #include "paddle/utils/TypeDefs.h" -#include "paddle/math/Vector.h" #include "ParameterConfig.pb.h" diff --git a/paddle/parameter/Parameter.cpp b/paddle/parameter/Parameter.cpp index 3b06650e0c..986ae1539b 100644 --- a/paddle/parameter/Parameter.cpp +++ b/paddle/parameter/Parameter.cpp @@ -12,19 +12,19 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "Parameter.h" #include -#include "paddle/math/MathUtils.h" #include "AverageOptimizer.h" #include "FirstOrderOptimizer.h" -#include "Parameter.h" -#include "paddle/utils/Logging.h" #include "OptimizerFunctions.h" #include "OptimizerWithRegularizer.h" #include "ParameterUpdateFunctions.h" -#include "paddle/math/SparseRowMatrix.h" -#include "paddle/math/CpuSparseMatrix.h" #include "hl_gpu.h" +#include "paddle/math/CpuSparseMatrix.h" +#include "paddle/math/MathUtils.h" +#include "paddle/math/SparseRowMatrix.h" #include "paddle/utils/CommandLineParser.h" +#include "paddle/utils/Logging.h" P_DEFINE_int32(enable_grad_share, (100 * 1024 * 1024), diff --git a/paddle/parameter/Parameter.h b/paddle/parameter/Parameter.h index 6b0600517a..532c6770e5 100644 --- a/paddle/parameter/Parameter.h +++ b/paddle/parameter/Parameter.h @@ -23,14 +23,14 @@ limitations under the License. */ #include "ParameterConfig.pb.h" #include "TrainerConfig.pb.h" +#include "ParameterUpdaterHook.h" +#include "paddle/math/Matrix.h" +#include "paddle/math/Vector.h" +#include "paddle/utils/GlobalConstants.h" #include "paddle/utils/Locks.h" +#include "paddle/utils/ThreadLocal.h" #include "paddle/utils/TypeDefs.h" -#include "paddle/math/Vector.h" -#include "paddle/math/Matrix.h" #include "paddle/utils/Util.h" -#include "paddle/utils/ThreadLocal.h" -#include "ParameterUpdaterHook.h" -#include "paddle/utils/GlobalConstants.h" namespace paddle { diff --git a/paddle/parameter/ParameterUpdateFunctions.h b/paddle/parameter/ParameterUpdateFunctions.h index 7374843d80..2d277e47e7 100644 --- a/paddle/parameter/ParameterUpdateFunctions.h +++ b/paddle/parameter/ParameterUpdateFunctions.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once -#include "paddle/utils/TypeDefs.h" #include "paddle/math/Vector.h" +#include "paddle/utils/TypeDefs.h" namespace paddle { diff --git a/paddle/parameter/ParameterUpdaterBase.cpp b/paddle/parameter/ParameterUpdaterBase.cpp index b938270ce1..49e2ae2b39 100644 --- a/paddle/parameter/ParameterUpdaterBase.cpp +++ b/paddle/parameter/ParameterUpdaterBase.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include -#include "paddle/utils/Logging.h" #include "ParameterUpdaterBase.h" +#include #include "hl_gpu.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/parameter/ParameterUpdaterHook.cpp b/paddle/parameter/ParameterUpdaterHook.cpp index 466560c437..f826e8448c 100644 --- a/paddle/parameter/ParameterUpdaterHook.cpp +++ b/paddle/parameter/ParameterUpdaterHook.cpp @@ -14,16 +14,16 @@ limitations under the License. */ #include "ParameterUpdaterHook.h" +#include #include -#include #include -#include #include +#include #include "paddle/math/Vector.h" #include "paddle/parameter/Parameter.h" -#include "paddle/utils/Util.h" #include "paddle/utils/Flags.h" +#include "paddle/utils/Util.h" namespace paddle { @@ -156,7 +156,8 @@ private: static WeakKVCache, IParameterUpdaterHook, - StringIntPairHasher> g_hookCache_; + StringIntPairHasher> + g_hookCache_; /** * ParameterUpdaterHook actually factory method. diff --git a/paddle/parameter/Regularizer.cpp b/paddle/parameter/Regularizer.cpp index 4420ee0031..8511900150 100644 --- a/paddle/parameter/Regularizer.cpp +++ b/paddle/parameter/Regularizer.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" -#include "paddle/utils/Flags.h" #include "Regularizer.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/parameter/Weight.cpp b/paddle/parameter/Weight.cpp index f366a2b53f..3738a58d7f 100644 --- a/paddle/parameter/Weight.cpp +++ b/paddle/parameter/Weight.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Logging.h" #include "Weight.h" +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/parameter/tests/test_common.cpp b/paddle/parameter/tests/test_common.cpp index 4e4d0ccfa2..aa57a63469 100644 --- a/paddle/parameter/tests/test_common.cpp +++ b/paddle/parameter/tests/test_common.cpp @@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include +#include #include -#include #include +#include #include #include diff --git a/paddle/pserver/BaseClient.cpp b/paddle/pserver/BaseClient.cpp index 62fafc1891..a43def98c5 100644 --- a/paddle/pserver/BaseClient.cpp +++ b/paddle/pserver/BaseClient.cpp @@ -12,11 +12,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include +#include "BaseClient.h" #include -#include "paddle/utils/Stat.h" +#include #include "paddle/utils/CommandLineParser.h" -#include "BaseClient.h" +#include "paddle/utils/Stat.h" P_DECLARE_string(pservers); diff --git a/paddle/pserver/BaseClient.h b/paddle/pserver/BaseClient.h index 5924f80684..262afafbe2 100644 --- a/paddle/pserver/BaseClient.h +++ b/paddle/pserver/BaseClient.h @@ -14,11 +14,11 @@ limitations under the License. */ #pragma once -#include "paddle/pserver/ProtoServer.h" +#include "ParameterService.pb.h" #include "paddle/math/Matrix.h" +#include "paddle/pserver/ProtoServer.h" #include "paddle/utils/Queue.h" #include "paddle/utils/TypeDefs.h" -#include "ParameterService.pb.h" namespace paddle { diff --git a/paddle/pserver/LightNetwork.cpp b/paddle/pserver/LightNetwork.cpp index 9a398d4f45..329dfb0fb3 100644 --- a/paddle/pserver/LightNetwork.cpp +++ b/paddle/pserver/LightNetwork.cpp @@ -12,23 +12,23 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include -#include +#include #include #include #include -#include +#include +#include #include -#include #include #include +#include #include #include "LightNetwork.h" -#include "paddle/utils/Util.h" -#include "paddle/utils/StringUtil.h" #include "RDMANetwork.h" +#include "paddle/utils/StringUtil.h" +#include "paddle/utils/Util.h" /// quick ack can reduce the latency of small message P_DEFINE_bool(small_messages, diff --git a/paddle/pserver/LightNetwork.h b/paddle/pserver/LightNetwork.h index 7aff007a27..c4a06deb94 100644 --- a/paddle/pserver/LightNetwork.h +++ b/paddle/pserver/LightNetwork.h @@ -16,10 +16,10 @@ limitations under the License. */ #include "SocketChannel.h" +#include #include #include #include -#include #include "paddle/utils/Thread.h" diff --git a/paddle/pserver/ParameterClient2.cpp b/paddle/pserver/ParameterClient2.cpp index 31418822b3..84d965a66a 100644 --- a/paddle/pserver/ParameterClient2.cpp +++ b/paddle/pserver/ParameterClient2.cpp @@ -15,10 +15,10 @@ limitations under the License. */ #include #include "ParameterClient2.h" -#include "paddle/utils/StringUtil.h" +#include "paddle/math/SparseRowMatrix.h" #include "paddle/utils/Flags.h" #include "paddle/utils/Stat.h" -#include "paddle/math/SparseRowMatrix.h" +#include "paddle/utils/StringUtil.h" P_DEFINE_string(pservers, "127.0.0.1", "Comma separated addresses of pservers"); P_DEFINE_int32(parallel_thread_num, 1, "Thread number for parameter send"); diff --git a/paddle/pserver/ParameterClient2.h b/paddle/pserver/ParameterClient2.h index 0f180722e3..5255394949 100644 --- a/paddle/pserver/ParameterClient2.h +++ b/paddle/pserver/ParameterClient2.h @@ -16,23 +16,23 @@ limitations under the License. */ #include #include -#include #include +#include -#include "paddle/utils/Locks.h" #include "paddle/math/Matrix.h" +#include "paddle/math/Vector.h" #include "paddle/parameter/Parameter.h" +#include "paddle/pserver/BaseClient.h" +#include "paddle/utils/Flags.h" +#include "paddle/utils/Locks.h" #include "paddle/utils/Queue.h" #include "paddle/utils/TypeDefs.h" #include "paddle/utils/Util.h" -#include "paddle/math/Vector.h" -#include "paddle/utils/Flags.h" -#include "paddle/pserver/BaseClient.h" #include "ParameterService.pb.h" -#include "SparseParameterDistribution.h" #include "ProtoServer.h" +#include "SparseParameterDistribution.h" P_DECLARE_int32(parallel_thread_num); diff --git a/paddle/pserver/ParameterServer2.cpp b/paddle/pserver/ParameterServer2.cpp index ac70efc64f..2cb4c93535 100644 --- a/paddle/pserver/ParameterServer2.cpp +++ b/paddle/pserver/ParameterServer2.cpp @@ -21,14 +21,14 @@ limitations under the License. */ #include "paddle/parameter/AverageOptimizer.h" #include "paddle/parameter/FirstOrderOptimizer.h" -#include "paddle/utils/Flags.h" #include "paddle/parameter/OptimizerFunctions.h" #include "paddle/parameter/OptimizerWithRegularizer.h" -#include "paddle/parameter/ParameterUpdateFunctions.h" #include "paddle/parameter/ParameterOptimizer.h" +#include "paddle/parameter/ParameterUpdateFunctions.h" #include "paddle/parameter/Regularizer.h" -#include "paddle/utils/Stat.h" +#include "paddle/utils/Flags.h" #include "paddle/utils/GlobalConstants.h" +#include "paddle/utils/Stat.h" P_DEFINE_int32(pserver_num_threads, 1, "number of threads for sync op exec"); P_DEFINE_double(async_lagged_ratio_min, diff --git a/paddle/pserver/ParameterServer2.h b/paddle/pserver/ParameterServer2.h index 47122f3632..61c139981e 100644 --- a/paddle/pserver/ParameterServer2.h +++ b/paddle/pserver/ParameterServer2.h @@ -15,24 +15,24 @@ limitations under the License. */ #pragma once #include +#include #include #include -#include -#include #include -#include +#include +#include #include #include -#include "paddle/utils/Locks.h" #include "paddle/math/Matrix.h" +#include "paddle/math/Vector.h" #include "paddle/parameter/Parameter.h" #include "paddle/parameter/ParameterOptimizer.h" +#include "paddle/utils/Locks.h" +#include "paddle/utils/Stat.h" #include "paddle/utils/ThreadLocal.h" #include "paddle/utils/TypeDefs.h" -#include "paddle/math/Vector.h" -#include "paddle/utils/Stat.h" #include "ParameterService.pb.h" diff --git a/paddle/pserver/ParameterServer2Main.cpp b/paddle/pserver/ParameterServer2Main.cpp index 1ba9b48c23..ffc521f2c1 100644 --- a/paddle/pserver/ParameterServer2Main.cpp +++ b/paddle/pserver/ParameterServer2Main.cpp @@ -12,13 +12,13 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" -#include "paddle/utils/StringUtil.h" #include +#include "paddle/utils/StringUtil.h" +#include "paddle/utils/Util.h" -#include "paddle/utils/Flags.h" #include "ParameterServer2.h" #include "RDMANetwork.h" +#include "paddle/utils/Flags.h" using namespace paddle; // NOLINT diff --git a/paddle/pserver/ProtoServer.h b/paddle/pserver/ProtoServer.h index 97b7bf167d..3acdcc27da 100644 --- a/paddle/pserver/ProtoServer.h +++ b/paddle/pserver/ProtoServer.h @@ -100,7 +100,8 @@ protected: ResponseCallback callback); typedef std::function msgReader, - ResponseCallback callback)> ServiceFunction; + ResponseCallback callback)> + ServiceFunction; /** * @brief register one RPC function in function mapping diff --git a/paddle/pserver/SocketChannel.cpp b/paddle/pserver/SocketChannel.cpp index f3e74257f6..0599889164 100644 --- a/paddle/pserver/SocketChannel.cpp +++ b/paddle/pserver/SocketChannel.cpp @@ -14,11 +14,11 @@ limitations under the License. */ #include "SocketChannel.h" -#include -#include -#include #include #include +#include +#include +#include #include #include "RDMANetwork.h" diff --git a/paddle/pserver/SparseParameterDistribution.h b/paddle/pserver/SparseParameterDistribution.h index dc63b065a7..24b14106cf 100644 --- a/paddle/pserver/SparseParameterDistribution.h +++ b/paddle/pserver/SparseParameterDistribution.h @@ -15,8 +15,8 @@ limitations under the License. */ #pragma once #include -#include "paddle/utils/Logging.h" #include +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/pserver/test/SocketTest.cpp b/paddle/pserver/test/SocketTest.cpp index 528f5e381e..6e63c4f678 100644 --- a/paddle/pserver/test/SocketTest.cpp +++ b/paddle/pserver/test/SocketTest.cpp @@ -14,11 +14,11 @@ limitations under the License. */ #include "paddle/utils/Util.h" -#include -#include -#include #include #include +#include +#include +#include #include diff --git a/paddle/pserver/test/test_ParameterServer2.cpp b/paddle/pserver/test/test_ParameterServer2.cpp index 493b6d060c..4257a2308d 100644 --- a/paddle/pserver/test/test_ParameterServer2.cpp +++ b/paddle/pserver/test/test_ParameterServer2.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include #include #include -#include #include #include diff --git a/paddle/pserver/test/test_ProtoServer.cpp b/paddle/pserver/test/test_ProtoServer.cpp index cfed0d30d3..3880dde5e3 100644 --- a/paddle/pserver/test/test_ProtoServer.cpp +++ b/paddle/pserver/test/test_ProtoServer.cpp @@ -16,10 +16,10 @@ limitations under the License. */ #include -#include "paddle/utils/Stat.h" +#include "ParameterService.pb.h" #include "paddle/math/Vector.h" #include "paddle/pserver/ProtoServer.h" -#include "ParameterService.pb.h" +#include "paddle/utils/Stat.h" P_DEFINE_string(server_addr, "127.0.0.1", "Server address"); P_DEFINE_int64(dim, 50000000, "Data size"); diff --git a/paddle/py_paddle/util.py b/paddle/py_paddle/util.py index d6bbf9a5a9..ce105d249a 100644 --- a/paddle/py_paddle/util.py +++ b/paddle/py_paddle/util.py @@ -559,10 +559,10 @@ def __monkey_patch_trainer__(): def monkeypatches(): - patches = [__monkeypatch_init_paddle__, - __monkeypatch_gradient_machine__, - __monkey_patch_protobuf_objects__, - __monkey_patch_parameter__, - __monkey_patch_trainer__] + patches = [ + __monkeypatch_init_paddle__, __monkeypatch_gradient_machine__, + __monkey_patch_protobuf_objects__, __monkey_patch_parameter__, + __monkey_patch_trainer__ + ] for patch in patches: patch() diff --git a/paddle/scripts/travis/main.sh b/paddle/scripts/travis/main.sh index c49d4546c2..13f2552d29 100755 --- a/paddle/scripts/travis/main.sh +++ b/paddle/scripts/travis/main.sh @@ -5,6 +5,8 @@ if [ ${JOB} == "BUILD_AND_TEST" ]; then ./build_and_test.sh elif [ ${JOB} == "DOCS" ]; then ./docs.sh +elif [ ${JOB} == "PRE_COMMIT" ]; then + ./precommit.sh else echo Unknown job ${JOB} exit 1 diff --git a/paddle/scripts/travis/precommit.sh b/paddle/scripts/travis/precommit.sh new file mode 100755 index 0000000000..3e70bc118e --- /dev/null +++ b/paddle/scripts/travis/precommit.sh @@ -0,0 +1,6 @@ +#!/bin/bash +set -e +source common.sh +cd .. +pre-commit install +pre-commit run -a diff --git a/paddle/trainer/MergeModel.cpp b/paddle/trainer/MergeModel.cpp index 8cb2873feb..1cf29a39b9 100644 --- a/paddle/trainer/MergeModel.cpp +++ b/paddle/trainer/MergeModel.cpp @@ -14,10 +14,10 @@ limitations under the License. */ #include -#include "paddle/utils/PythonUtil.h" -#include "paddle/pserver/ParameterServer2.h" #include "ParamUtil.h" #include "Trainer.h" +#include "paddle/pserver/ParameterServer2.h" +#include "paddle/utils/PythonUtil.h" P_DEFINE_string(model_dir, "", "Directory for separated model files"); P_DEFINE_string(model_file, "", "File for merged model file"); diff --git a/paddle/trainer/ParamUtil.cpp b/paddle/trainer/ParamUtil.cpp index 200417ebfc..ffbca42e10 100644 --- a/paddle/trainer/ParamUtil.cpp +++ b/paddle/trainer/ParamUtil.cpp @@ -17,22 +17,22 @@ limitations under the License. */ #include #include -#include #include -#include +#include #include +#include #include #include +#include "paddle/utils/GlobalConstants.h" #include "paddle/utils/PythonUtil.h" #include "paddle/utils/Stat.h" #include "paddle/utils/Util.h" -#include "paddle/utils/GlobalConstants.h" +#include "TesterConfig.h" #include "paddle/gserver/gradientmachines/NeuralNetwork.h" #include "paddle/gserver/layers/ValidationLayer.h" -#include "TesterConfig.h" namespace paddle { diff --git a/paddle/trainer/ParamUtil.h b/paddle/trainer/ParamUtil.h index 8fa6fda75c..2e05595848 100644 --- a/paddle/trainer/ParamUtil.h +++ b/paddle/trainer/ParamUtil.h @@ -22,11 +22,11 @@ limitations under the License. */ #include "paddle/gserver/dataproviders/DataProvider.h" #include "paddle/gserver/gradientmachines/GradientMachine.h" +#include +#include +#include "ParameterUpdater.h" #include "TrainerConfig.pb.h" #include "TrainerConfigHelper.h" -#include "ParameterUpdater.h" -#include -#include namespace paddle { diff --git a/paddle/trainer/ParameterUpdater.h b/paddle/trainer/ParameterUpdater.h index 81ac374425..e52b5cd318 100644 --- a/paddle/trainer/ParameterUpdater.h +++ b/paddle/trainer/ParameterUpdater.h @@ -24,8 +24,8 @@ limitations under the License. */ #include "paddle/parameter/Parameter.h" #include "paddle/parameter/ParameterUpdaterBase.h" -#include "paddle/gserver/layers/Layer.h" #include "TrainerConfig.pb.h" +#include "paddle/gserver/layers/Layer.h" #include #include diff --git a/paddle/trainer/RemoteParameterUpdater.cpp b/paddle/trainer/RemoteParameterUpdater.cpp index 702ea07f8a..b7f7b93b8d 100644 --- a/paddle/trainer/RemoteParameterUpdater.cpp +++ b/paddle/trainer/RemoteParameterUpdater.cpp @@ -14,8 +14,8 @@ limitations under the License. */ #include "RemoteParameterUpdater.h" #include "Trainer.h" -#include "paddle/utils/Stat.h" #include "paddle/utils/GlobalConstants.h" +#include "paddle/utils/Stat.h" P_DECLARE_int32(trainer_id); P_DECLARE_string(save_dir); diff --git a/paddle/trainer/RemoteParameterUpdater.h b/paddle/trainer/RemoteParameterUpdater.h index 46ce4be146..66055c778e 100644 --- a/paddle/trainer/RemoteParameterUpdater.h +++ b/paddle/trainer/RemoteParameterUpdater.h @@ -14,12 +14,12 @@ limitations under the License. */ #pragma once -#include #include -#include "paddle/pserver/ParameterClient2.h" +#include #include "ParameterUpdater.h" -#include "paddle/utils/Util.h" +#include "paddle/pserver/ParameterClient2.h" #include "paddle/utils/Queue.h" +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/trainer/Tester.h b/paddle/trainer/Tester.h index ae7e0e93bf..e892744db2 100644 --- a/paddle/trainer/Tester.h +++ b/paddle/trainer/Tester.h @@ -24,12 +24,12 @@ limitations under the License. */ #include "TrainerConfig.pb.h" -#include "ParameterUpdater.h" +#include +#include #include "ParamUtil.h" +#include "ParameterUpdater.h" #include "TesterConfig.h" #include "TrainerInternalConfig.h" -#include -#include namespace paddle { diff --git a/paddle/trainer/TesterConfig.h b/paddle/trainer/TesterConfig.h index 9ff145a8a1..68d4c931ff 100644 --- a/paddle/trainer/TesterConfig.h +++ b/paddle/trainer/TesterConfig.h @@ -23,9 +23,9 @@ limitations under the License. */ #include "TrainerConfig.pb.h" -#include "ParameterUpdater.h" -#include #include +#include +#include "ParameterUpdater.h" namespace paddle { diff --git a/paddle/trainer/ThreadParameterUpdater.h b/paddle/trainer/ThreadParameterUpdater.h index 492692dbe5..d01ac689f9 100644 --- a/paddle/trainer/ThreadParameterUpdater.h +++ b/paddle/trainer/ThreadParameterUpdater.h @@ -14,13 +14,13 @@ limitations under the License. */ #pragma once -#include "paddle/utils/Util.h" #include "paddle/parameter/AverageOptimizer.h" #include "paddle/parameter/FirstOrderOptimizer.h" #include "paddle/parameter/OptimizerFunctions.h" #include "paddle/parameter/OptimizerWithRegularizer.h" #include "paddle/parameter/Parameter.h" #include "paddle/parameter/Regularizer.h" +#include "paddle/utils/Util.h" #include #include diff --git a/paddle/trainer/Trainer.h b/paddle/trainer/Trainer.h index f50b56143d..cabbb4acd1 100644 --- a/paddle/trainer/Trainer.h +++ b/paddle/trainer/Trainer.h @@ -22,13 +22,13 @@ limitations under the License. */ #include "paddle/gserver/dataproviders/DataProvider.h" #include "paddle/gserver/gradientmachines/GradientMachine.h" -#include "TrainerConfigHelper.h" +#include +#include +#include "ParamUtil.h" #include "ParameterUpdater.h" -#include "TrainerInternal.h" #include "Tester.h" -#include "ParamUtil.h" -#include -#include +#include "TrainerConfigHelper.h" +#include "TrainerInternal.h" #ifdef PADDLE_METRIC_LEARNING #include "paddle/internals/metric_learning/MetricTrainer.h" diff --git a/paddle/trainer/TrainerConfigHelper.h b/paddle/trainer/TrainerConfigHelper.h index 2c5c492ce8..f1366cc041 100644 --- a/paddle/trainer/TrainerConfigHelper.h +++ b/paddle/trainer/TrainerConfigHelper.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once -#include #include #include +#include namespace paddle { diff --git a/paddle/trainer/TrainerInternal.cpp b/paddle/trainer/TrainerInternal.cpp index 1b49d4aa28..f3b465b444 100644 --- a/paddle/trainer/TrainerInternal.cpp +++ b/paddle/trainer/TrainerInternal.cpp @@ -17,22 +17,22 @@ limitations under the License. */ #include #include -#include #include -#include +#include #include +#include #include +#include "paddle/gserver/gradientmachines/NeuralNetwork.h" +#include "paddle/gserver/layers/ValidationLayer.h" +#include "paddle/utils/GlobalConstants.h" #include "paddle/utils/PythonUtil.h" #include "paddle/utils/Stat.h" #include "paddle/utils/Util.h" -#include "paddle/utils/GlobalConstants.h" -#include "paddle/gserver/gradientmachines/NeuralNetwork.h" -#include "paddle/gserver/layers/ValidationLayer.h" -#include "ThreadParameterUpdater.h" #include "RemoteParameterUpdater.h" +#include "ThreadParameterUpdater.h" namespace paddle { diff --git a/paddle/trainer/TrainerInternal.h b/paddle/trainer/TrainerInternal.h index b67711a721..7018faab24 100644 --- a/paddle/trainer/TrainerInternal.h +++ b/paddle/trainer/TrainerInternal.h @@ -17,15 +17,15 @@ limitations under the License. */ #include "paddle/utils/Util.h" #include -#include #include +#include -#include "hl_gpu.h" -#include "paddle/gserver/gradientmachines/GradientMachine.h" -#include "TrainerConfig.pb.h" #include "ParameterUpdater.h" +#include "TrainerConfig.pb.h" #include "TrainerConfigHelper.h" #include "TrainerInternalConfig.h" +#include "hl_gpu.h" +#include "paddle/gserver/gradientmachines/GradientMachine.h" namespace paddle { diff --git a/paddle/trainer/TrainerInternalConfig.h b/paddle/trainer/TrainerInternalConfig.h index fd6fdf45e6..b47692720e 100644 --- a/paddle/trainer/TrainerInternalConfig.h +++ b/paddle/trainer/TrainerInternalConfig.h @@ -23,10 +23,10 @@ limitations under the License. */ #include "TrainerConfig.pb.h" -#include "ParameterUpdater.h" +#include #include #include -#include +#include "ParameterUpdater.h" namespace paddle { /** diff --git a/paddle/trainer/TrainerMain.cpp b/paddle/trainer/TrainerMain.cpp index 7a18f9836c..0a4d56b892 100644 --- a/paddle/trainer/TrainerMain.cpp +++ b/paddle/trainer/TrainerMain.cpp @@ -13,10 +13,10 @@ See the License for the specific language governing permissions and limitations under the License. */ #include +#include "paddle/pserver/ParameterServer2.h" +#include "paddle/utils/Excepts.h" #include "paddle/utils/PythonUtil.h" #include "paddle/utils/StringUtil.h" -#include "paddle/utils/Excepts.h" -#include "paddle/pserver/ParameterServer2.h" #include "ParamUtil.h" #include "Trainer.h" diff --git a/paddle/trainer/tests/picojson.h b/paddle/trainer/tests/picojson.h index cb657d219e..23bfa16408 100644 --- a/paddle/trainer/tests/picojson.h +++ b/paddle/trainer/tests/picojson.h @@ -30,10 +30,10 @@ #define picojson_h #include +#include #include #include #include -#include #include #include #include diff --git a/paddle/trainer/tests/test_Compare.cpp b/paddle/trainer/tests/test_Compare.cpp index 07a47b2990..63fa48540c 100644 --- a/paddle/trainer/tests/test_Compare.cpp +++ b/paddle/trainer/tests/test_Compare.cpp @@ -16,8 +16,8 @@ limitations under the License. */ #include "paddle/trainer/Trainer.h" -#include #include +#include using namespace paddle; // NOLINT using namespace std; // NOLINT diff --git a/paddle/trainer/tests/test_CompareTwoNets.cpp b/paddle/trainer/tests/test_CompareTwoNets.cpp index 7e5449dcba..8a4556721d 100644 --- a/paddle/trainer/tests/test_CompareTwoNets.cpp +++ b/paddle/trainer/tests/test_CompareTwoNets.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include #include -#include #include -#include +#include #include "paddle/trainer/Trainer.h" diff --git a/paddle/trainer/tests/test_CompareTwoOpts.cpp b/paddle/trainer/tests/test_CompareTwoOpts.cpp index 4d051b537c..673ef289d8 100644 --- a/paddle/trainer/tests/test_CompareTwoOpts.cpp +++ b/paddle/trainer/tests/test_CompareTwoOpts.cpp @@ -12,10 +12,10 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include #include -#include #include -#include +#include #include "paddle/trainer/Trainer.h" diff --git a/paddle/trainer/tests/test_PyDataProviderWrapper.cpp b/paddle/trainer/tests/test_PyDataProviderWrapper.cpp index 5c5c6d5346..66ec65e340 100644 --- a/paddle/trainer/tests/test_PyDataProviderWrapper.cpp +++ b/paddle/trainer/tests/test_PyDataProviderWrapper.cpp @@ -13,16 +13,16 @@ See the License for the specific language governing permissions and limitations under the License. */ #ifndef PADDLE_NO_PYTHON +#include #include -#include #include -#include #include #include +#include +#include +#include #include #include -#include -#include #include "picojson.h" void checkEqual(const paddle::Argument& expect, const paddle::Argument& actual); diff --git a/paddle/trainer/tests/test_TrainerOnePass.cpp b/paddle/trainer/tests/test_TrainerOnePass.cpp index 1d9dce1b0e..0b587ecce1 100644 --- a/paddle/trainer/tests/test_TrainerOnePass.cpp +++ b/paddle/trainer/tests/test_TrainerOnePass.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include #include +#include #include "paddle/trainer/Trainer.h" #include "paddle/trainer/TrainerInternal.h" diff --git a/paddle/trainer/tests/test_recurrent_machine_generation.cpp b/paddle/trainer/tests/test_recurrent_machine_generation.cpp index b52acc2ca7..7d8dfd788f 100644 --- a/paddle/trainer/tests/test_recurrent_machine_generation.cpp +++ b/paddle/trainer/tests/test_recurrent_machine_generation.cpp @@ -14,8 +14,8 @@ limitations under the License. */ #include -#include #include +#include #include diff --git a/paddle/utils/BarrierStat.cpp b/paddle/utils/BarrierStat.cpp index 5040deefd0..9dde155aca 100644 --- a/paddle/utils/BarrierStat.cpp +++ b/paddle/utils/BarrierStat.cpp @@ -12,13 +12,13 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "paddle/utils/BarrierStat.h" +#include #include -#include #include -#include -#include "paddle/utils/Stat.h" -#include "paddle/utils/BarrierStat.h" +#include #include "paddle/utils/Flags.h" +#include "paddle/utils/Stat.h" P_DEFINE_bool(log_barrier_abstract, true, diff --git a/paddle/utils/BarrierStat.h b/paddle/utils/BarrierStat.h index 3c5c0885d6..352f641b6e 100644 --- a/paddle/utils/BarrierStat.h +++ b/paddle/utils/BarrierStat.h @@ -15,18 +15,18 @@ limitations under the License. */ #pragma once #include -#include #include -#include #include +#include +#include #include +#include #include -#include -#include "Logging.h" #include "Locks.h" -#include "ThreadLocal.h" +#include "Logging.h" #include "Stat.h" +#include "ThreadLocal.h" namespace paddle { diff --git a/paddle/utils/CommandLineParser.cpp b/paddle/utils/CommandLineParser.cpp index 14f83241c5..51558b45a1 100644 --- a/paddle/utils/CommandLineParser.cpp +++ b/paddle/utils/CommandLineParser.cpp @@ -14,15 +14,15 @@ limitations under the License. */ #include "CommandLineParser.h" #ifndef PADDLE_USE_GFLAGS -#include "paddle/utils/StringUtil.h" +#include #include -#include #include -#include +#include #include -#include -#include #include +#include +#include +#include "paddle/utils/StringUtil.h" namespace paddle { @@ -46,16 +46,13 @@ template <> bool StringToValue(const std::string& content, bool* value) { std::string tmp = content; - std::transform(tmp.begin(), - tmp.end(), - tmp.begin(), - [](char in) -> char { - if (in <= 'Z' && in >= 'A') { - return in - ('Z' - 'z'); - } else { - return in; - } - }); // tolower. + std::transform(tmp.begin(), tmp.end(), tmp.begin(), [](char in) -> char { + if (in <= 'Z' && in >= 'A') { + return in - ('Z' - 'z'); + } else { + return in; + } + }); // tolower. if (tmp == "true" || tmp == "1") { *value = true; diff --git a/paddle/utils/CommandLineParser.h b/paddle/utils/CommandLineParser.h index 3d25bc3b0b..b4449c6f09 100644 --- a/paddle/utils/CommandLineParser.h +++ b/paddle/utils/CommandLineParser.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once #ifndef PADDLE_USE_GFLAGS -#include "DisableCopy.h" +#include #include #include -#include +#include "DisableCopy.h" namespace paddle { diff --git a/paddle/utils/CpuId.cpp b/paddle/utils/CpuId.cpp index 734b2e0924..53db82e48a 100644 --- a/paddle/utils/CpuId.cpp +++ b/paddle/utils/CpuId.cpp @@ -15,43 +15,43 @@ limitations under the License. */ #ifdef _WIN32 /// for MSVC -#define CPUID(info, x) __cpuidex(info, x, 0) +#define CPUID(info, x) __cpuidex(info, x, 0) #else #include /// for GCC/Clang -#define CPUID(info, x) __cpuid_count(x, 0, info[0], info[1], info[2], info[3]) +#define CPUID(info, x) __cpuid_count(x, 0, info[0], info[1], info[2], info[3]) #endif namespace paddle { SIMDFlags::SIMDFlags() { - unsigned int cpuInfo[4]; - // CPUID: https://en.wikipedia.org/wiki/CPUID - CPUID(cpuInfo, 0x00000001); - simd_flags_ |= cpuInfo[3] & (1 << 25) ? SIMD_SSE : SIMD_NONE; - simd_flags_ |= cpuInfo[3] & (1 << 26) ? SIMD_SSE2 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 0) ? SIMD_SSE3 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 9) ? SIMD_SSSE3 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 19) ? SIMD_SSE41 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 20) ? SIMD_SSE42 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 12) ? SIMD_FMA3 : SIMD_NONE; - simd_flags_ |= cpuInfo[2] & (1 << 28) ? SIMD_AVX : SIMD_NONE; - - CPUID(cpuInfo, 0x00000007); - simd_flags_ |= cpuInfo[1] & (1 << 5) ? SIMD_AVX2 : SIMD_NONE; - simd_flags_ |= cpuInfo[1] & (1 << 16) ? SIMD_AVX512: SIMD_NONE; - - CPUID(cpuInfo, 0x80000001); - simd_flags_ |= cpuInfo[2] & (1 << 16) ? SIMD_FMA4 : SIMD_NONE; + unsigned int cpuInfo[4]; + // CPUID: https://en.wikipedia.org/wiki/CPUID + CPUID(cpuInfo, 0x00000001); + simd_flags_ |= cpuInfo[3] & (1 << 25) ? SIMD_SSE : SIMD_NONE; + simd_flags_ |= cpuInfo[3] & (1 << 26) ? SIMD_SSE2 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 0) ? SIMD_SSE3 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 9) ? SIMD_SSSE3 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 19) ? SIMD_SSE41 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 20) ? SIMD_SSE42 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 12) ? SIMD_FMA3 : SIMD_NONE; + simd_flags_ |= cpuInfo[2] & (1 << 28) ? SIMD_AVX : SIMD_NONE; + + CPUID(cpuInfo, 0x00000007); + simd_flags_ |= cpuInfo[1] & (1 << 5) ? SIMD_AVX2 : SIMD_NONE; + simd_flags_ |= cpuInfo[1] & (1 << 16) ? SIMD_AVX512 : SIMD_NONE; + + CPUID(cpuInfo, 0x80000001); + simd_flags_ |= cpuInfo[2] & (1 << 16) ? SIMD_FMA4 : SIMD_NONE; } SIMDFlags* SIMDFlags::instance() { - static SIMDFlags instance; - return &instance; + static SIMDFlags instance; + return &instance; } -} // namespace paddle +} // namespace paddle diff --git a/paddle/utils/CpuId.h b/paddle/utils/CpuId.h index d15e58d1dd..66ac59cf3e 100644 --- a/paddle/utils/CpuId.h +++ b/paddle/utils/CpuId.h @@ -18,54 +18,54 @@ namespace paddle { class SIMDFlags final { public: - DISABLE_COPY(SIMDFlags); + DISABLE_COPY(SIMDFlags); - SIMDFlags(); + SIMDFlags(); - static SIMDFlags* instance(); + static SIMDFlags* instance(); - inline bool isSSE() const { return simd_flags_ & SIMD_SSE; } - inline bool isSSE2() const { return simd_flags_ & SIMD_SSE2; } - inline bool isSSE3() const { return simd_flags_ & SIMD_SSE3; } - inline bool isSSSE3() const { return simd_flags_ & SIMD_SSSE3; } - inline bool isSSE41() const { return simd_flags_ & SIMD_SSE41; } - inline bool isSSE42() const { return simd_flags_ & SIMD_SSE42; } - inline bool isFMA3() const { return simd_flags_ & SIMD_FMA3; } - inline bool isFMA4() const { return simd_flags_ & SIMD_FMA4; } - inline bool isAVX() const { return simd_flags_ & SIMD_AVX; } - inline bool isAVX2() const { return simd_flags_ & SIMD_AVX2; } - inline bool isAVX512()const { return simd_flags_ & SIMD_AVX512;} + inline bool isSSE() const { return simd_flags_ & SIMD_SSE; } + inline bool isSSE2() const { return simd_flags_ & SIMD_SSE2; } + inline bool isSSE3() const { return simd_flags_ & SIMD_SSE3; } + inline bool isSSSE3() const { return simd_flags_ & SIMD_SSSE3; } + inline bool isSSE41() const { return simd_flags_ & SIMD_SSE41; } + inline bool isSSE42() const { return simd_flags_ & SIMD_SSE42; } + inline bool isFMA3() const { return simd_flags_ & SIMD_FMA3; } + inline bool isFMA4() const { return simd_flags_ & SIMD_FMA4; } + inline bool isAVX() const { return simd_flags_ & SIMD_AVX; } + inline bool isAVX2() const { return simd_flags_ & SIMD_AVX2; } + inline bool isAVX512() const { return simd_flags_ & SIMD_AVX512; } private: - enum simd_t { - SIMD_NONE = 0, ///< None - SIMD_SSE = 1 << 0, ///< SSE - SIMD_SSE2 = 1 << 1, ///< SSE 2 - SIMD_SSE3 = 1 << 2, ///< SSE 3 - SIMD_SSSE3 = 1 << 3, ///< SSSE 3 - SIMD_SSE41 = 1 << 4, ///< SSE 4.1 - SIMD_SSE42 = 1 << 5, ///< SSE 4.2 - SIMD_FMA3 = 1 << 6, ///< FMA 3 - SIMD_FMA4 = 1 << 7, ///< FMA 4 - SIMD_AVX = 1 << 8, ///< AVX - SIMD_AVX2 = 1 << 9, ///< AVX 2 - SIMD_AVX512 = 1 << 10, ///< AVX 512 - }; + enum simd_t { + SIMD_NONE = 0, ///< None + SIMD_SSE = 1 << 0, ///< SSE + SIMD_SSE2 = 1 << 1, ///< SSE 2 + SIMD_SSE3 = 1 << 2, ///< SSE 3 + SIMD_SSSE3 = 1 << 3, ///< SSSE 3 + SIMD_SSE41 = 1 << 4, ///< SSE 4.1 + SIMD_SSE42 = 1 << 5, ///< SSE 4.2 + SIMD_FMA3 = 1 << 6, ///< FMA 3 + SIMD_FMA4 = 1 << 7, ///< FMA 4 + SIMD_AVX = 1 << 8, ///< AVX + SIMD_AVX2 = 1 << 9, ///< AVX 2 + SIMD_AVX512 = 1 << 10, ///< AVX 512 + }; - /// simd flags - int simd_flags_ = SIMD_NONE; + /// simd flags + int simd_flags_ = SIMD_NONE; }; -#define HAS_SSE SIMDFlags::instance()->isSSE() -#define HAS_SSE2 SIMDFlags::instance()->isSSE2() -#define HAS_SSE3 SIMDFlags::instance()->isSSE3() -#define HAS_SSSE3 SIMDFlags::instance()->isSSSE3() -#define HAS_SSE41 SIMDFlags::instance()->isSSE41() -#define HAS_SSE42 SIMDFlags::instance()->isSSE42() -#define HAS_FMA3 SIMDFlags::instance()->isFMA3() -#define HAS_FMA4 SIMDFlags::instance()->isFMA4() -#define HAS_AVX SIMDFlags::instance()->isAVX() -#define HAS_AVX2 SIMDFlags::instance()->isAVX2() -#define HAS_AVX512 SIMDFlags::instance()->isAVX512() +#define HAS_SSE SIMDFlags::instance()->isSSE() +#define HAS_SSE2 SIMDFlags::instance()->isSSE2() +#define HAS_SSE3 SIMDFlags::instance()->isSSE3() +#define HAS_SSSE3 SIMDFlags::instance()->isSSSE3() +#define HAS_SSE41 SIMDFlags::instance()->isSSE41() +#define HAS_SSE42 SIMDFlags::instance()->isSSE42() +#define HAS_FMA3 SIMDFlags::instance()->isFMA3() +#define HAS_FMA4 SIMDFlags::instance()->isFMA4() +#define HAS_AVX SIMDFlags::instance()->isAVX() +#define HAS_AVX2 SIMDFlags::instance()->isAVX2() +#define HAS_AVX512 SIMDFlags::instance()->isAVX512() -} // namespace paddle +} // namespace paddle diff --git a/paddle/utils/CustomStackTrace.cpp b/paddle/utils/CustomStackTrace.cpp index 730788cb98..083f5c509a 100644 --- a/paddle/utils/CustomStackTrace.cpp +++ b/paddle/utils/CustomStackTrace.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "CustomStackTrace.h" -#include "CommandLineParser.h" #include +#include "CommandLineParser.h" P_DEFINE_bool( layer_stack_error_only_current_thread, diff --git a/paddle/utils/CustomStackTrace.h b/paddle/utils/CustomStackTrace.h index 5686f3c84c..6992e85622 100644 --- a/paddle/utils/CustomStackTrace.h +++ b/paddle/utils/CustomStackTrace.h @@ -14,10 +14,10 @@ limitations under the License. */ #pragma once +#include #include #include #include -#include #include "ThreadLocal.h" @@ -96,7 +96,8 @@ public: */ typedef std::function DumpCallback; + const T& /*item*/)> + DumpCallback; /** * Dump all thread stack, and all stack will be cleared. diff --git a/paddle/utils/Logging.cpp b/paddle/utils/Logging.cpp index 3c31633e58..20f32466a5 100644 --- a/paddle/utils/Logging.cpp +++ b/paddle/utils/Logging.cpp @@ -22,13 +22,13 @@ limitations under the License. */ #include #include #include -#include -#include #include +#include +#include -#include -#include #include +#include +#include #include namespace paddle { diff --git a/paddle/utils/Logging.h b/paddle/utils/Logging.h index c91ca9fecc..4379289f6d 100644 --- a/paddle/utils/Logging.h +++ b/paddle/utils/Logging.h @@ -18,8 +18,8 @@ limitations under the License. */ */ #pragma once -#include #include +#include #include #ifndef PADDLE_USE_GLOG diff --git a/paddle/utils/PythonUtil.cpp b/paddle/utils/PythonUtil.cpp index a9c6a20997..2ee4e4fb7e 100644 --- a/paddle/utils/PythonUtil.cpp +++ b/paddle/utils/PythonUtil.cpp @@ -13,8 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "PythonUtil.h" -#include #include +#include namespace paddle { diff --git a/paddle/utils/PythonUtil.h b/paddle/utils/PythonUtil.h index 2cbc2fdd37..9e2a1c360c 100644 --- a/paddle/utils/PythonUtil.h +++ b/paddle/utils/PythonUtil.h @@ -36,10 +36,10 @@ limitations under the License. */ #endif -#include "paddle/utils/Util.h" #include -#include #include +#include +#include "paddle/utils/Util.h" namespace paddle { diff --git a/paddle/utils/Queue.h b/paddle/utils/Queue.h index 37748345a4..f054738f87 100644 --- a/paddle/utils/Queue.h +++ b/paddle/utils/Queue.h @@ -142,9 +142,9 @@ public: */ bool waitNotEmptyFor(int seconds) { std::unique_lock lock(queueLock_); - return queueCV_.wait_for(lock, - std::chrono::seconds(seconds), - [this] { return numElements_ != 0; }); + return queueCV_.wait_for(lock, std::chrono::seconds(seconds), [this] { + return numElements_ != 0; + }); } private: diff --git a/paddle/utils/Stat.cpp b/paddle/utils/Stat.cpp index 01ea535cfd..44acee2495 100644 --- a/paddle/utils/Stat.cpp +++ b/paddle/utils/Stat.cpp @@ -13,9 +13,9 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "Stat.h" -#include "Util.h" -#include #include +#include +#include "Util.h" namespace paddle { @@ -207,10 +207,9 @@ static unsigned g_profileCount = 0; static std::recursive_mutex g_profileMutex; GpuProfiler::GpuProfiler(std::string statName, std::string info) - : guard_(g_profileMutex) { + : guard_(g_profileMutex) { if (++g_profileCount == 1) { - LOG(INFO) << "Enable GPU Profiler Stat: [" - << statName << "] " << info; + LOG(INFO) << "Enable GPU Profiler Stat: [" << statName << "] " << info; hl_profiler_start(); } } diff --git a/paddle/utils/StringUtil.h b/paddle/utils/StringUtil.h index 8a63ca23b4..0b4f4c9113 100644 --- a/paddle/utils/StringUtil.h +++ b/paddle/utils/StringUtil.h @@ -14,9 +14,9 @@ limitations under the License. */ #pragma once +#include #include #include -#include #include "Logging.h" namespace paddle { diff --git a/paddle/utils/Thread.h b/paddle/utils/Thread.h index 435dff2f66..ef36a8c5b2 100644 --- a/paddle/utils/Thread.h +++ b/paddle/utils/Thread.h @@ -13,9 +13,9 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once -#include "Util.h" -#include "Logging.h" #include +#include "Logging.h" +#include "Util.h" #include "Queue.h" #include "ThreadLocal.h" diff --git a/paddle/utils/ThreadLocal.cpp b/paddle/utils/ThreadLocal.cpp index c9b32784d9..8a2878fc4b 100644 --- a/paddle/utils/ThreadLocal.cpp +++ b/paddle/utils/ThreadLocal.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "Util.h" #include "ThreadLocal.h" #include "CommandLineParser.h" +#include "Util.h" P_DEFINE_bool(thread_local_rand_use_global_seed, false, diff --git a/paddle/utils/ThreadLocal.h b/paddle/utils/ThreadLocal.h index b6e31bd05b..a4987c9ec2 100644 --- a/paddle/utils/ThreadLocal.h +++ b/paddle/utils/ThreadLocal.h @@ -15,14 +15,14 @@ limitations under the License. */ #pragma once #include -#include #include +#include #include #include #include #include -#include "Util.h" #include "Logging.h" +#include "Util.h" namespace paddle { diff --git a/paddle/utils/Util.cpp b/paddle/utils/Util.cpp index f48726bff0..26ff385c84 100644 --- a/paddle/utils/Util.cpp +++ b/paddle/utils/Util.cpp @@ -15,11 +15,11 @@ limitations under the License. */ #include "Util.h" #include +#include #include #include #include #include -#include #include #include @@ -28,10 +28,10 @@ limitations under the License. */ #include "CommandLineParser.h" #include "CustomStackTrace.h" +#include "StringUtil.h" #include "Thread.h" #include "ThreadLocal.h" #include "Version.h" -#include "StringUtil.h" P_DEFINE_int32(seed, 1, "random number seed. 0 for srand(time)"); @@ -126,25 +126,23 @@ void registerInitFunction(std::function func, int priority) { } void runInitFunctions() { - std::call_once( - g_onceFlag, - []() { - LOG(INFO) << "Calling runInitFunctions"; - if (g_initFuncs) { - std::sort(g_initFuncs->begin(), - g_initFuncs->end(), - [](const PriorityFuncPair& x, const PriorityFuncPair& y) { - return x.first > y.first; - }); - for (auto& f : *g_initFuncs) { - f.second(); - } - delete g_initFuncs; - g_initFuncs = nullptr; - } - g_initialized = true; - LOG(INFO) << "Call runInitFunctions done."; - }); + std::call_once(g_onceFlag, []() { + LOG(INFO) << "Calling runInitFunctions"; + if (g_initFuncs) { + std::sort(g_initFuncs->begin(), + g_initFuncs->end(), + [](const PriorityFuncPair& x, const PriorityFuncPair& y) { + return x.first > y.first; + }); + for (auto& f : *g_initFuncs) { + f.second(); + } + delete g_initFuncs; + g_initFuncs = nullptr; + } + g_initialized = true; + LOG(INFO) << "Call runInitFunctions done."; + }); } void initMain(int argc, char** argv) { diff --git a/paddle/utils/Util.h b/paddle/utils/Util.h index ff67439da6..24ddde28e7 100644 --- a/paddle/utils/Util.h +++ b/paddle/utils/Util.h @@ -14,25 +14,25 @@ limitations under the License. */ #pragma once +#include // for syscall() +#include #include #include -#include -#include +#include #include +#include +#include #include #include -#include -#include -#include // for syscall() -#include +#include #include "CommandLineParser.h" +#include "DisableCopy.h" #include "Logging.h" #include "TrainerConfig.pb.h" -#include "DisableCopy.h" -#include "TypeDefs.h" #include "Flags.h" +#include "TypeDefs.h" #include "hl_gpu.h" /** diff --git a/paddle/utils/Version.cpp b/paddle/utils/Version.cpp index 086515791d..a9e351b69f 100644 --- a/paddle/utils/Version.cpp +++ b/paddle/utils/Version.cpp @@ -14,10 +14,10 @@ limitations under the License. */ #include "Version.h" -#include "Flags.h" -#include "Util.h" #include #include +#include "Flags.h" +#include "Util.h" //! TODO(yuyang18) in gflags, version has another define. Use another flag //! instead. #ifndef PADDLE_USE_GFLAGS @@ -33,7 +33,8 @@ void printVersion(std::ostream& os) { #ifndef PADDLE_VERSION #define PADDLE_VERSION "unknown" #endif -// converts macro to string https://gcc.gnu.org/onlinedocs/cpp/Stringification.html +// converts macro to string +// https://gcc.gnu.org/onlinedocs/cpp/Stringification.html #define xstr(s) str(s) #define str(s) #s diff --git a/paddle/utils/Version.h b/paddle/utils/Version.h index ac04963c2c..d1a07d9485 100644 --- a/paddle/utils/Version.h +++ b/paddle/utils/Version.h @@ -14,8 +14,8 @@ limitations under the License. */ #pragma once #include -#include "TypeDefs.h" #include +#include "TypeDefs.h" namespace paddle { diff --git a/paddle/utils/arch/osx/Locks.cpp b/paddle/utils/arch/osx/Locks.cpp index 8590226431..e03992363f 100644 --- a/paddle/utils/arch/osx/Locks.cpp +++ b/paddle/utils/arch/osx/Locks.cpp @@ -13,10 +13,10 @@ See the License for the specific language governing permissions and limitations under the License. */ #include "paddle/utils/Locks.h" -#include "paddle/utils/Logging.h" #include -#include #include +#include +#include "paddle/utils/Logging.h" namespace paddle { diff --git a/paddle/utils/tests/test_CommandLineParser.cpp b/paddle/utils/tests/test_CommandLineParser.cpp index 9a1d2391a8..ed2b3068d5 100644 --- a/paddle/utils/tests/test_CommandLineParser.cpp +++ b/paddle/utils/tests/test_CommandLineParser.cpp @@ -15,8 +15,8 @@ limitations under the License. */ #ifndef PADDLE_USE_GFLAGS //! Test Command Line Parser for paddle internal implement. -#include #include +#include P_DEFINE_int32(i1, 1, "test int flag 1"); P_DEFINE_int32(i2, 2, "test int flag 2"); diff --git a/paddle/utils/tests/test_CustomStackTrace.cpp b/paddle/utils/tests/test_CustomStackTrace.cpp index 512330b49e..292ed4619d 100644 --- a/paddle/utils/tests/test_CustomStackTrace.cpp +++ b/paddle/utils/tests/test_CustomStackTrace.cpp @@ -15,10 +15,10 @@ limitations under the License. */ #include #include -#include "paddle/utils/CustomStackTrace.h" #include "paddle/utils/CommandLineParser.h" -#include "paddle/utils/Util.h" +#include "paddle/utils/CustomStackTrace.h" #include "paddle/utils/Locks.h" +#include "paddle/utils/Util.h" P_DEFINE_int32(test_thread_num, 10, "testing thread number"); diff --git a/paddle/utils/tests/test_CustomStackTracePrint.cpp b/paddle/utils/tests/test_CustomStackTracePrint.cpp index 60ba210b70..611b16aa71 100644 --- a/paddle/utils/tests/test_CustomStackTracePrint.cpp +++ b/paddle/utils/tests/test_CustomStackTracePrint.cpp @@ -12,8 +12,8 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include "paddle/utils/Util.h" #include "paddle/utils/CustomStackTrace.h" +#include "paddle/utils/Util.h" int main(int argc, char** argv) { paddle::initMain(argc, argv); diff --git a/paddle/utils/tests/test_Logging.cpp b/paddle/utils/tests/test_Logging.cpp index 667864aa75..fbfffcc65a 100644 --- a/paddle/utils/tests/test_Logging.cpp +++ b/paddle/utils/tests/test_Logging.cpp @@ -17,10 +17,10 @@ limitations under the License. */ * Used in embedded system where there is no glogs. */ +#include #include -#include #include -#include +#include #include "paddle/utils/Logging.h" #include "paddle/utils/Util.h" #ifndef PADDLE_USE_GLOG diff --git a/paddle/utils/tests/test_SIMDFlags.cpp b/paddle/utils/tests/test_SIMDFlags.cpp index a544901aa3..41532953a7 100644 --- a/paddle/utils/tests/test_SIMDFlags.cpp +++ b/paddle/utils/tests/test_SIMDFlags.cpp @@ -9,40 +9,39 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ - #include #include "paddle/utils/CpuId.h" #include "paddle/utils/Logging.h" #include "paddle/utils/Util.h" -using namespace paddle; // NOLINT +using namespace paddle; // NOLINT TEST(SIMDFlags, gccTest) { #if (defined(__GNUC__) || defined(__GNUG__)) && !(defined(__clang__)) - CHECK(!__builtin_cpu_supports("sse") != HAS_SSE); - CHECK(!__builtin_cpu_supports("sse2") != HAS_SSE2); - CHECK(!__builtin_cpu_supports("sse3") != HAS_SSE3); - CHECK(!__builtin_cpu_supports("ssse3") != HAS_SSSE3); - CHECK(!__builtin_cpu_supports("sse4.1")!= HAS_SSE41); - CHECK(!__builtin_cpu_supports("sse4.2")!= HAS_SSE42); - CHECK(!__builtin_cpu_supports("avx") != HAS_AVX); - CHECK(!__builtin_cpu_supports("avx2") != HAS_AVX2); + CHECK(!__builtin_cpu_supports("sse") != HAS_SSE); + CHECK(!__builtin_cpu_supports("sse2") != HAS_SSE2); + CHECK(!__builtin_cpu_supports("sse3") != HAS_SSE3); + CHECK(!__builtin_cpu_supports("ssse3") != HAS_SSSE3); + CHECK(!__builtin_cpu_supports("sse4.1") != HAS_SSE41); + CHECK(!__builtin_cpu_supports("sse4.2") != HAS_SSE42); + CHECK(!__builtin_cpu_supports("avx") != HAS_AVX); + CHECK(!__builtin_cpu_supports("avx2") != HAS_AVX2); #endif } TEST(SIMDFlags, normalPrint) { - auto simd = SIMDFlags::instance(); - LOG(INFO) << "Has SSE2: " << std::boolalpha << simd->isSSE2(); - LOG(INFO) << "Has SSE3: " << std::boolalpha << simd->isSSE3(); - LOG(INFO) << "Has SSSE3: " << std::boolalpha << simd->isSSSE3(); - LOG(INFO) << "Has SSE4.1: " << std::boolalpha << simd->isSSE41(); - LOG(INFO) << "Has SSE4.2: " << std::boolalpha << simd->isSSE42(); - LOG(INFO) << "Has FMA3: " << std::boolalpha << simd->isFMA3(); - LOG(INFO) << "Has FMA4: " << std::boolalpha << simd->isFMA4(); - LOG(INFO) << "Has AVX: " << std::boolalpha << simd->isAVX(); - LOG(INFO) << "Has AVX2: " << std::boolalpha << simd->isAVX2(); - LOG(INFO) << "Has AVX512: " << std::boolalpha << simd->isAVX512(); + auto simd = SIMDFlags::instance(); + LOG(INFO) << "Has SSE2: " << std::boolalpha << simd->isSSE2(); + LOG(INFO) << "Has SSE3: " << std::boolalpha << simd->isSSE3(); + LOG(INFO) << "Has SSSE3: " << std::boolalpha << simd->isSSSE3(); + LOG(INFO) << "Has SSE4.1: " << std::boolalpha << simd->isSSE41(); + LOG(INFO) << "Has SSE4.2: " << std::boolalpha << simd->isSSE42(); + LOG(INFO) << "Has FMA3: " << std::boolalpha << simd->isFMA3(); + LOG(INFO) << "Has FMA4: " << std::boolalpha << simd->isFMA4(); + LOG(INFO) << "Has AVX: " << std::boolalpha << simd->isAVX(); + LOG(INFO) << "Has AVX2: " << std::boolalpha << simd->isAVX2(); + LOG(INFO) << "Has AVX512: " << std::boolalpha << simd->isAVX512(); } int main(int argc, char** argv) { diff --git a/paddle/utils/tests/test_SpinLock.cpp b/paddle/utils/tests/test_SpinLock.cpp index 9c7ad05b0b..22f8584ef5 100644 --- a/paddle/utils/tests/test_SpinLock.cpp +++ b/paddle/utils/tests/test_SpinLock.cpp @@ -14,10 +14,10 @@ limitations under the License. */ #include #include -#include "paddle/utils/Logging.h" #include "paddle/utils/CommandLineParser.h" -#include "paddle/utils/Util.h" #include "paddle/utils/Locks.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/Util.h" P_DEFINE_int32(test_thread_num, 100, "testing thread number"); diff --git a/paddle/utils/tests/test_Thread.cpp b/paddle/utils/tests/test_Thread.cpp index b069be1d7a..2f5c5bbce0 100644 --- a/paddle/utils/tests/test_Thread.cpp +++ b/paddle/utils/tests/test_Thread.cpp @@ -12,9 +12,9 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ -#include -#include #include +#include +#include using paddle::AsyncThreadPool; // NOLINT @@ -52,17 +52,13 @@ TEST(AsyncThreadPool, multiThreadAddBatchJob) { int counter = 0; const int numMonitors = 300; const int numSlaves = 300; - std::vector moniterJobs( - numMonitors, - [&] { - std::vector slaveJobs( - numSlaves, - [mut, &counter] { - std::lock_guard lk(*mut); - counter++; - }); - levelTwoPool.addBatchJobs(slaveJobs); - }); + std::vector moniterJobs(numMonitors, [&] { + std::vector slaveJobs(numSlaves, [mut, &counter] { + std::lock_guard lk(*mut); + counter++; + }); + levelTwoPool.addBatchJobs(slaveJobs); + }); levelOnePool.addBatchJobs(moniterJobs); ASSERT_EQ(counter, numMonitors * numSlaves); } diff --git a/paddle/utils/tests/test_ThreadBarrier.cpp b/paddle/utils/tests/test_ThreadBarrier.cpp index 997a393683..4a8af5b97e 100644 --- a/paddle/utils/tests/test_ThreadBarrier.cpp +++ b/paddle/utils/tests/test_ThreadBarrier.cpp @@ -15,10 +15,10 @@ limitations under the License. */ #include #include #include -#include "paddle/utils/Logging.h" #include "paddle/utils/CommandLineParser.h" -#include "paddle/utils/Util.h" #include "paddle/utils/Locks.h" +#include "paddle/utils/Logging.h" +#include "paddle/utils/Util.h" P_DEFINE_int32(test_thread_num, 100, "testing thread number"); From 3aa00440b1864badfe612b6bf241f896c829cc3e Mon Sep 17 00:00:00 2001 From: Yi Wang Date: Sun, 11 Dec 2016 23:02:47 -0800 Subject: [PATCH 56/60] Fix bugs in Docker document --- .../build_and_install/docker_install_en.rst | 26 ++++++++++++------- 1 file changed, 17 insertions(+), 9 deletions(-) diff --git a/doc/getstarted/build_and_install/docker_install_en.rst b/doc/getstarted/build_and_install/docker_install_en.rst index feb027ccbb..8df7e063a1 100644 --- a/doc/getstarted/build_and_install/docker_install_en.rst +++ b/doc/getstarted/build_and_install/docker_install_en.rst @@ -122,9 +122,9 @@ The general development workflow with Docker and Bazel is as follows: git clone --recursive https://github.com/paddlepaddle/paddle -2. Build a development Docker image `paddle:dev` from the source code. - This image contains all the development tools and dependencies of - PaddlePaddle. +2. Build a development Docker image :code:`paddle:dev` from the source + code. This image contains all the development tools and + dependencies of PaddlePaddle. .. code-block:: bash @@ -139,14 +139,22 @@ The general development workflow with Docker and Bazel is as follows: .. code-block:: bash - docker run \ - -d # run the container in background mode \ - --name paddle # we can run a nginx container to serve documents \ - -p 2022:22 # so we can SSH into this container \ - -v $PWD:/paddle # mount the source code \ - -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \ + docker run \ + -d \ + --name paddle \ + -p 2022:22 \ + -v $PWD:/paddle \ + -v $HOME/.cache/bazel:/root/.cache/bazel \ paddle:dev + where :code:`-d` makes the container running in background, + :code:`--name paddle` allows us to run a nginx container to serve + documents in this container, :code:`-p 2022:22` allows us to SSH + into this container, :code:`-v $PWD:/paddle` shares the source code + on the host with the container, :code:`-v + $HOME/.cache/bazel:/root/.cache/bazel` shares Bazel cache on the + host with the container. + 4. SSH into the container: .. code-block:: bash From be1b70e64e719bb1f1af5ea414e9e1aaf8f0a89b Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 15:08:55 +0800 Subject: [PATCH 57/60] Tuning travis --- paddle/cuda/src/hl_cuda_device.cc | 2 +- paddle/scripts/travis/precommit.sh | 2 ++ 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/paddle/cuda/src/hl_cuda_device.cc b/paddle/cuda/src/hl_cuda_device.cc index d181448292..b0bba73594 100644 --- a/paddle/cuda/src/hl_cuda_device.cc +++ b/paddle/cuda/src/hl_cuda_device.cc @@ -12,13 +12,13 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ +#include "hl_cuda.h" #include #include #include #include #include #include -#include "hl_cuda.h" #include "hl_cuda.ph" #include "hl_dso_loader.h" #include "hl_thread.ph" diff --git a/paddle/scripts/travis/precommit.sh b/paddle/scripts/travis/precommit.sh index 3e70bc118e..48d2f437a1 100755 --- a/paddle/scripts/travis/precommit.sh +++ b/paddle/scripts/travis/precommit.sh @@ -2,5 +2,7 @@ set -e source common.sh cd .. +export PATH=/usr/bin:$PATH pre-commit install +clang-format --version pre-commit run -a From 4afaaa4b9210a851e735bf79aadb92420d773477 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 15:33:41 +0800 Subject: [PATCH 58/60] Autoformat all files --- WORKSPACE | 20 ++++----- demo/sentiment/predict.py | 11 +++-- python/paddle/trainer/config_parser.py | 4 ++ .../default_decorators.py | 3 ++ .../test_config_parser_for_non_file_config.py | 44 ++++++++++--------- .../tests/test_reset_hook.py | 10 ++--- third_party/gtest.BUILD | 20 +++------ third_party/protobuf_test/BUILD | 29 ++++++------ third_party/protobuf_test/example_lib.cc | 4 +- 9 files changed, 70 insertions(+), 75 deletions(-) diff --git a/WORKSPACE b/WORKSPACE index d6ae2af8eb..0b8299905a 100644 --- a/WORKSPACE +++ b/WORKSPACE @@ -1,17 +1,15 @@ # External dependency to Google protobuf. http_archive( - name = "protobuf", - url = "http://github.com/google/protobuf/archive/v3.1.0.tar.gz", - sha256 = "0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7", - strip_prefix = "protobuf-3.1.0", -) + name="protobuf", + url="http://github.com/google/protobuf/archive/v3.1.0.tar.gz", + sha256="0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7", + strip_prefix="protobuf-3.1.0", ) # External dependency to gtest 1.7.0. This method comes from # https://www.bazel.io/versions/master/docs/tutorial/cpp.html. new_http_archive( - name = "gtest", - url = "https://github.com/google/googletest/archive/release-1.7.0.zip", - sha256 = "b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0", - build_file = "third_party/gtest.BUILD", - strip_prefix = "googletest-release-1.7.0", -) + name="gtest", + url="https://github.com/google/googletest/archive/release-1.7.0.zip", + sha256="b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0", + build_file="third_party/gtest.BUILD", + strip_prefix="googletest-release-1.7.0", ) diff --git a/demo/sentiment/predict.py b/demo/sentiment/predict.py index 0095c6f727..8ec490f646 100755 --- a/demo/sentiment/predict.py +++ b/demo/sentiment/predict.py @@ -71,9 +71,7 @@ class SentimentPrediction(): transform word into integer index according to the dictionary. """ words = data.strip().split() - word_slot = [ - self.word_dict[w] for w in words if w in self.word_dict - ] + word_slot = [self.word_dict[w] for w in words if w in self.word_dict] return word_slot def batch_predict(self, data_batch): @@ -85,8 +83,8 @@ class SentimentPrediction(): if self.label is None: print("predicting label is %d" % (lab[0])) else: - print("predicting label is %s" % - (self.label[lab[0]])) + print("predicting label is %s" % (self.label[lab[0]])) + def option_parser(): usage = "python predict.py -n config -w model_dir -d dictionary -i input_file " @@ -143,9 +141,10 @@ def main(): batch.append([predict.get_index(line)]) if len(batch) == batch_size: predict.batch_predict(batch) - batch=[] + batch = [] if len(batch) > 0: predict.batch_predict(batch) + if __name__ == '__main__': main() diff --git a/python/paddle/trainer/config_parser.py b/python/paddle/trainer/config_parser.py index 42a7a29403..5b7f4d85e2 100644 --- a/python/paddle/trainer/config_parser.py +++ b/python/paddle/trainer/config_parser.py @@ -3364,7 +3364,10 @@ def my_fatal(s): logger.critical(s) raise Exception() + _parse_config_hooks = set() + + def register_parse_config_hook(f): """ Register a hook function for parse_config. parse_config will invoke the hook @@ -3373,6 +3376,7 @@ def register_parse_config_hook(f): """ _parse_config_hooks.add(f) + def parse_config(config_file, config_arg_str): ''' @param config_arg_str: a string of the form var1=val1,var2=val2. It will be diff --git a/python/paddle/trainer_config_helpers/default_decorators.py b/python/paddle/trainer_config_helpers/default_decorators.py index 13712aad7b..ad3efcbf36 100644 --- a/python/paddle/trainer_config_helpers/default_decorators.py +++ b/python/paddle/trainer_config_helpers/default_decorators.py @@ -84,12 +84,15 @@ class DefaultNameFactory(object): _name_factories = [] + def reset_hook(): for factory in _name_factories: factory.reset() + register_parse_config_hook(reset_hook) + def wrap_name_default(name_prefix=None): """ Decorator to set "name" arguments default to "{name_prefix}_{invoke_count}". diff --git a/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py index 87a607acf4..9b791a0222 100644 --- a/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py +++ b/python/paddle/trainer_config_helpers/tests/configs/test_config_parser_for_non_file_config.py @@ -17,33 +17,35 @@ import sys import re import getopt + def main(print_whole_config, globals, locals): - ''' + ''' this test will all test_config.py ''' - cmdstr = """from paddle.trainer.config_parser import parse_config\n""" - importstr = "" - functionstr = "" + cmdstr = """from paddle.trainer.config_parser import parse_config\n""" + importstr = "" + functionstr = "" + + for line in sys.stdin: + if re.match("^import", line) or re.match("^from.*import", line): + importstr = importstr + line + else: + functionstr = functionstr + " " + line - for line in sys.stdin: - if re.match("^import", line) or re.match("^from.*import", line): - importstr = importstr + line + cmdstr = cmdstr + importstr + """def configs():\n""" + functionstr + #cmdstr = cmdstr + """def configs():\n""" + importstr + functionstr + if print_whole_config: + cmdstr = cmdstr + """print parse_config(configs, "")""" else: - functionstr = functionstr + " " + line + cmdstr = cmdstr + """print parse_config(configs, "").model_config""" - cmdstr = cmdstr + importstr + """def configs():\n""" + functionstr - #cmdstr = cmdstr + """def configs():\n""" + importstr + functionstr - if print_whole_config: - cmdstr = cmdstr + """print parse_config(configs, "")""" - else: - cmdstr = cmdstr + """print parse_config(configs, "").model_config""" + exec (cmdstr, globals, locals) - exec(cmdstr, globals, locals) if __name__ == '__main__': - whole = False - opts, args = getopt.getopt(sys.argv[1:], "", ["whole"]) - for op, value in opts: - if op == "--whole": - whole = True - main(whole, globals(), locals()) + whole = False + opts, args = getopt.getopt(sys.argv[1:], "", ["whole"]) + for op, value in opts: + if op == "--whole": + whole = True + main(whole, globals(), locals()) diff --git a/python/paddle/trainer_config_helpers/tests/test_reset_hook.py b/python/paddle/trainer_config_helpers/tests/test_reset_hook.py index dc494d0eef..0423babdb7 100644 --- a/python/paddle/trainer_config_helpers/tests/test_reset_hook.py +++ b/python/paddle/trainer_config_helpers/tests/test_reset_hook.py @@ -14,13 +14,13 @@ import unittest from paddle.trainer.config_parser import parse_config -class TestParse(unittest.TestCase): +class TestParse(unittest.TestCase): def test_parse(self): - a = parse_config( - 'trainer_config_helpers/tests/layers_test_config.py', '') - b = parse_config( - 'trainer_config_helpers/tests/layers_test_config.py', '') + a = parse_config('trainer_config_helpers/tests/layers_test_config.py', + '') + b = parse_config('trainer_config_helpers/tests/layers_test_config.py', + '') self.assertEqual(a, b) diff --git a/third_party/gtest.BUILD b/third_party/gtest.BUILD index 3e68a1d879..71c74af513 100644 --- a/third_party/gtest.BUILD +++ b/third_party/gtest.BUILD @@ -1,14 +1,8 @@ cc_library( - name = "main", - srcs = glob( - ["src/*.cc"], - exclude = ["src/gtest-all.cc"] - ), - hdrs = glob([ - "include/**/*.h", - "src/*.h" - ]), - copts = ["-Iexternal/gtest/include"], - linkopts = ["-pthread"], - visibility = ["//visibility:public"], -) + name="main", + srcs=glob( + ["src/*.cc"], exclude=["src/gtest-all.cc"]), + hdrs=glob(["include/**/*.h", "src/*.h"]), + copts=["-Iexternal/gtest/include"], + linkopts=["-pthread"], + visibility=["//visibility:public"], ) diff --git a/third_party/protobuf_test/BUILD b/third_party/protobuf_test/BUILD index 46f769da5f..95a687a356 100644 --- a/third_party/protobuf_test/BUILD +++ b/third_party/protobuf_test/BUILD @@ -3,25 +3,22 @@ licenses(["notice"]) # Apache 2.0 load("@protobuf//:protobuf.bzl", "cc_proto_library") cc_proto_library( - name = "example_proto", - srcs = ["example.proto"], - protoc = "@protobuf//:protoc", - default_runtime = "@protobuf//:protobuf", -) + name="example_proto", + srcs=["example.proto"], + protoc="@protobuf//:protoc", + default_runtime="@protobuf//:protobuf", ) cc_library( - name = "example_lib", - srcs = ["example_lib.cc"], - hdrs = ["example_lib.h"], - deps = [":example_proto"], -) + name="example_lib", + srcs=["example_lib.cc"], + hdrs=["example_lib.h"], + deps=[":example_proto"], ) cc_test( - name = "example_lib_test", - srcs = ["example_lib_test.cc"], - copts = ["-Iexternal/gtest/include"], - deps =[ + name="example_lib_test", + srcs=["example_lib_test.cc"], + copts=["-Iexternal/gtest/include"], + deps=[ "@gtest//:main", ":example_lib", - ], -) + ], ) diff --git a/third_party/protobuf_test/example_lib.cc b/third_party/protobuf_test/example_lib.cc index 56341a0124..ced377bc0a 100644 --- a/third_party/protobuf_test/example_lib.cc +++ b/third_party/protobuf_test/example_lib.cc @@ -3,9 +3,7 @@ namespace third_party { namespace protobuf_test { -std::string get_greet(const Greeting& who) { - return "Hello " + who.name(); -} +std::string get_greet(const Greeting& who) { return "Hello " + who.name(); } } // namespace protobuf_test } // namespace thrid_party From 6ea5a9fd0361d58bcea205bc9736166cf17e1b94 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 15:43:06 +0800 Subject: [PATCH 59/60] Add tips when unittest error --- paddle/scripts/travis/precommit.sh | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/paddle/scripts/travis/precommit.sh b/paddle/scripts/travis/precommit.sh index 48d2f437a1..5ad84f1821 100755 --- a/paddle/scripts/travis/precommit.sh +++ b/paddle/scripts/travis/precommit.sh @@ -1,4 +1,11 @@ #!/bin/bash +function abort(){ + echo "Your commit not fit PaddlePaddle code style" 1>&2 + echo "Please use pre-commit scripts to auto-format your code" 1>&2 + exit 1 +} + +trap 'abort' 0 set -e source common.sh cd .. @@ -6,3 +13,5 @@ export PATH=/usr/bin:$PATH pre-commit install clang-format --version pre-commit run -a + +trap : 0 From 579e59120708b3fb0190347da1dab5189dfd23d3 Mon Sep 17 00:00:00 2001 From: Yu Yang Date: Mon, 12 Dec 2016 15:45:28 +0800 Subject: [PATCH 60/60] Try to fix unittest error --- paddle/cuda/include/hl_time.h | 2 +- paddle/cuda/src/hl_time.cc | 1 + paddle/utils/BarrierStat.h | 1 - paddle/utils/PythonUtil.h | 5 +++-- 4 files changed, 5 insertions(+), 4 deletions(-) diff --git a/paddle/cuda/include/hl_time.h b/paddle/cuda/include/hl_time.h index f214b055f9..f63f025820 100644 --- a/paddle/cuda/include/hl_time.h +++ b/paddle/cuda/include/hl_time.h @@ -14,7 +14,7 @@ limitations under the License. */ #ifndef HL_TIME_H_ #define HL_TIME_H_ - +#include /** * @brief High resolution timer. * diff --git a/paddle/cuda/src/hl_time.cc b/paddle/cuda/src/hl_time.cc index 2bb69d25e5..7e5d7e8aae 100644 --- a/paddle/cuda/src/hl_time.cc +++ b/paddle/cuda/src/hl_time.cc @@ -15,6 +15,7 @@ limitations under the License. */ #include "hl_time.h" #include #include +#include #include using std::chrono::high_resolution_clock; diff --git a/paddle/utils/BarrierStat.h b/paddle/utils/BarrierStat.h index 352f641b6e..a9c925eff6 100644 --- a/paddle/utils/BarrierStat.h +++ b/paddle/utils/BarrierStat.h @@ -25,7 +25,6 @@ limitations under the License. */ #include "Locks.h" #include "Logging.h" -#include "Stat.h" #include "ThreadLocal.h" namespace paddle { diff --git a/paddle/utils/PythonUtil.h b/paddle/utils/PythonUtil.h index 9e2a1c360c..daebaffc85 100644 --- a/paddle/utils/PythonUtil.h +++ b/paddle/utils/PythonUtil.h @@ -13,6 +13,8 @@ See the License for the specific language governing permissions and limitations under the License. */ #pragma once +// clang-format off +#include "paddle/utils/Util.h" #ifndef PADDLE_NO_PYTHON // must include the following two blocks, otherwise, @@ -33,13 +35,12 @@ limitations under the License. */ #endif #include #include - #endif #include #include #include -#include "paddle/utils/Util.h" +// clang-format on namespace paddle {