m53297601
/
Paddle
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Polish save load en doc details (#27845)

Browse Source 
* polish save load en doc details, test=document_fix

* polish details, test=document_fix

* remove blank, test=document_fix

* fix ascii error, test=document_fix

* update fc's input to x, test=document_fix

* update fc argument activation, test=document_fix
my_2.0rc
Chen Weihang 4 years ago committed by GitHub
parent 92b3a71705
commit 94faa112e9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 38 additions and 33 deletions
Show Stats Download Patch File Download Diff File

6
python/paddle/fluid/dygraph/checkpoint.py
Unescape View File

@ -140,12 +140,12 @@ def load_dygraph(model_path, **configs):
Args: Args:
model_path(str) : The file prefix store the state_dict. model_path(str) : The file prefix store the state_dict.
(The path should Not contain suffix '.pdparams') (The path should Not contain suffix '.pdparams')
**configs (dict, optional): other save configuration options for compatibility. We do not **configs (dict, optional): Other load configuration options for compatibility. We do not
recommend using these configurations, if not necessary, DO NOT use them. Default None. recommend using these configurations, if not necessary, DO NOT use them. Default None.
The following options are currently supported: The following options are currently supported:
(1) model_filename (string): The inference model file name of the paddle 1.x ``save_inference_model`` (1) model_filename (str): The inference model file name of the paddle 1.x ``save_inference_model``
save format. Default file name is :code:`__model__` . save format. Default file name is :code:`__model__` .
(2) params_filename (string): The persistable variables file name of the paddle 1.x ``save_inference_model`` (2) params_filename (str): The persistable variables file name of the paddle 1.x ``save_inference_model``
save format. No default file name, save variables separately by default. save format. No default file name, save variables separately by default.
Returns: Returns:

32
python/paddle/fluid/dygraph/jit.py
Unescape View File

@ -470,10 +470,10 @@ def save(layer, path, input_spec=None, **configs):
format model, which can be used for inference or fine-tuning after loading. format model, which can be used for inference or fine-tuning after loading.
It will save the translated program and all related persistable It will save the translated program and all related persistable
variables of input Layer to given ``path``. variables of input Layer to given ``path`` .
``path`` is the prefix of saved objects, and the saved translated program file ``path`` is the prefix of saved objects, and the saved translated program file
suffix is ``.pdmodel``, the saved persistable variables file suffix is ``.pdiparams``, suffix is ``.pdmodel`` , the saved persistable variables file suffix is ``.pdiparams`` ,
and here also saved some additional variable description information to a file, and here also saved some additional variable description information to a file,
its suffix is ``.pdiparams.info``, these additional information is used in fine-tuning. its suffix is ``.pdiparams.info``, these additional information is used in fine-tuning.
@ -483,18 +483,17 @@ def save(layer, path, input_spec=None, **configs):
- Other C++ inference APIs - Other C++ inference APIs
Args: Args:
layer (Layer): the Layer to be saved. The Layer should be decorated by `@paddle.jit.to_static`. layer (Layer): The Layer to be saved.
path (str): The path prefix to save model. The format is ``dirname/file_prefix`` or ``file_prefix``. path (str): The path prefix to save model. The format is ``dirname/file_prefix`` or ``file_prefix``.
input_spec (list[InputSpec|Tensor], optional): Describes the input of the saved model. input_spec (list[InputSpec|Tensor], optional): Describes the input of the saved model's forward
It is the example inputs that will be passed to saved TranslatedLayer's forward method, which can be described by InputSpec or example Tensor. If None, all input variables of
function. If None, all input variables of the original Layer's forward function the original Layer's forward method would be the inputs of the saved model. Default None.
would be the inputs of the saved model. Default None. **configs (dict, optional): Other save configuration options for compatibility. We do not
**configs (dict, optional): other save configuration options for compatibility. We do not
recommend using these configurations, they may be removed in the future. If not necessary, recommend using these configurations, they may be removed in the future. If not necessary,
DO NOT use them. Default None. DO NOT use them. Default None.
The following options are currently supported: The following options are currently supported:
(1) output_spec (list[Tensor]): Selects the output targets of the saved model. (1) output_spec (list[Tensor]): Selects the output targets of the saved model.
By default, all return variables of original Layer's forward function are kept as the By default, all return variables of original Layer's forward method are kept as the
output of the saved model. If the provided ``output_spec`` list is not all output variables, output of the saved model. If the provided ``output_spec`` list is not all output variables,
the saved model will be pruned according to the given ``output_spec`` list. the saved model will be pruned according to the given ``output_spec`` list.
@ -735,14 +734,14 @@ def load(path, **configs):
4. The parameter's ``trainable`` information is lost and can not be recovered. 4. The parameter's ``trainable`` information is lost and can not be recovered.
Args: Args:
path (str): The path prefix to load model. The format is ``dirname/file_prefix`` or ``file_prefix``. path (str): The path prefix to load model. The format is ``dirname/file_prefix`` or ``file_prefix`` .
**configs (dict, optional): other load configuration options for compatibility. We do not **configs (dict, optional): Other load configuration options for compatibility. We do not
recommend using these configurations, they may be removed in the future. If not necessary, recommend using these configurations, they may be removed in the future. If not necessary,
DO NOT use them. Default None. DO NOT use them. Default None.
The following options are currently supported: The following options are currently supported:
(1) model_filename (string): The inference model file name of the paddle 1.x (1) model_filename (str): The inference model file name of the paddle 1.x
``save_inference_model`` save format. Default file name is :code:`__model__` . ``save_inference_model`` save format. Default file name is :code:`__model__` .
(2) params_filename (string): The persistable variables file name of the paddle 1.x (2) params_filename (str): The persistable variables file name of the paddle 1.x
``save_inference_model`` save format. No default file name, save variables separately ``save_inference_model`` save format. No default file name, save variables separately
by default. by default.
@ -844,7 +843,6 @@ def load(path, **configs):
import numpy as np import numpy as np
import paddle import paddle
import paddle.fluid as fluid
import paddle.static as static import paddle.static as static
import paddle.nn as nn import paddle.nn as nn
import paddle.optimizer as opt import paddle.optimizer as opt
@ -870,9 +868,11 @@ def load(path, **configs):
def __len__(self): def __len__(self):
return self.num_samples return self.num_samples
paddle.enable_static()
image = static.data(name='image', shape=[None, 784], dtype='float32') image = static.data(name='image', shape=[None, 784], dtype='float32')
label = static.data(name='label', shape=[None, 1], dtype='int64') label = static.data(name='label', shape=[None, 1], dtype='int64')
pred = static.nn.fc(input=image, size=10, act='softmax') pred = static.nn.fc(x=image, size=10, activation='softmax')
loss = F.cross_entropy(input=pred, label=label) loss = F.cross_entropy(input=pred, label=label)
avg_loss = paddle.mean(loss) avg_loss = paddle.mean(loss)
@ -901,7 +901,7 @@ def load(path, **configs):
fetch_list=[avg_loss]) fetch_list=[avg_loss])
model_path = "fc.example.model" model_path = "fc.example.model"
fluid.io.save_inference_model( paddle.fluid.io.save_inference_model(
model_path, ["image"], [pred], exe) model_path, ["image"], [pred], exe)
# 2. load model # 2. load model

33
python/paddle/framework/io.py
Unescape View File

@ -204,10 +204,13 @@ def save(obj, path):
Now only supports save ``state_dict`` of Layer or Optimizer. Now only supports save ``state_dict`` of Layer or Optimizer.
.. note:: .. note::
``paddle.save`` will not add a suffix to the saved results, Different from ``paddle.jit.save``, since the save result of ``paddle.save`` is a single file,
but we recommend that you use the following paddle standard suffixes: there is no need to distinguish multiple saved files by adding a suffix. The argument ``path``
1. for ``Layer.state_dict`` -> ``.pdparams`` of ``paddle.save`` will be directly used as the saved file name instead of a prefix.
2. for ``Optimizer.state_dict`` -> ``.pdopt`` In order to unify the saved file name format, we recommend using the paddle standard suffix:
1. for ``Layer.state_dict`` , recommend to use ``.pdparams`` ;
2. for ``Optimizer.state_dict`` , recommend to use ``.pdopt`` .
For specific examples, please refer to API code examples.
Args: Args:
obj(Object) : The object to be saved. obj(Object) : The object to be saved.
@ -272,9 +275,10 @@ def load(path, **configs):
Now only supports load ``state_dict`` of Layer or Optimizer. Now only supports load ``state_dict`` of Layer or Optimizer.
.. note:: .. note::
``paddle.load`` supports loading ``state_dict`` of Layer or Optimizer from In order to use the model parameters saved by paddle more efficiently,
the result of other save APIs except ``paddle.load`` , but the argument ``paddle.load`` supports loading ``state_dict`` of Layer from the result of
``path`` format is different: other save APIs except ``paddle.save`` , but the argument ``path`` format is
different:
1. loading from ``paddle.static.save`` or ``paddle.Model().save(training=True)`` , 1. loading from ``paddle.static.save`` or ``paddle.Model().save(training=True)`` ,
``path`` needs to be a complete file name, such as ``model.pdparams`` or ``path`` needs to be a complete file name, such as ``model.pdparams`` or
``model.pdopt`` ; ``model.pdopt`` ;
@ -287,22 +291,23 @@ def load(path, **configs):
directory, such as ``model`` and model is a directory. directory, such as ``model`` and model is a directory.
.. note:: .. note::
If you load ``state_dict`` from the saved result of If you load ``state_dict`` from the saved result of static mode API such as
``paddle.static.save`` or ``paddle.static.save_inference_model`` , ``paddle.static.save`` or ``paddle.static.save_inference_model`` ,
the structured variable name will cannot be restored. You need to set the argument the structured variable name in dynamic mode will cannot be restored.
``use_structured_name=False`` when using ``Layer.set_state_dict`` later. You need to set the argument ``use_structured_name=False`` when using
``Layer.set_state_dict`` later.
Args: Args:
path(str) : The path to load the target object. Generally, the path is the target path(str) : The path to load the target object. Generally, the path is the target
file path. When compatible with loading the saved results other APIs, the path file path. When loading state_dict from the saved result of the API used to save
can be a file prefix or directory. the inference model, the path may be a file prefix or directory.
**configs (dict, optional): other load configuration options for compatibility. We do not **configs (dict, optional): other load configuration options for compatibility. We do not
recommend using these configurations, they may be removed in the future. If not necessary, recommend using these configurations, they may be removed in the future. If not necessary,
DO NOT use them. Default None. DO NOT use them. Default None.
The following options are currently supported: The following options are currently supported:
(1) model_filename (string): The inference model file name of the paddle 1.x (1) model_filename (str): The inference model file name of the paddle 1.x
``save_inference_model`` save format. Default file name is :code:`__model__` . ``save_inference_model`` save format. Default file name is :code:`__model__` .
(2) params_filename (string): The persistable variables file name of the paddle 1.x (2) params_filename (str): The persistable variables file name of the paddle 1.x
``save_inference_model`` save format. No default file name, save variables separately ``save_inference_model`` save format. No default file name, save variables separately
by default. by default.

Write Preview
Loading…
Cancel
Save