| .. |
|
benchmark
|
Optimize the CUDA implementation of sequence_expand op by reduce the times of copying lod data from CPU to GPU. (#15493)
|
6 years ago |
|
controlflow
|
Revert "Optimize while_op when is_test is true. (#15811)" (#15968)
|
6 years ago |
|
csp
|
Refine operator cmake (#14413)
|
6 years ago |
|
detail
|
Merge pull request #14933 from sneaxiy/rewrite_ddim
|
6 years ago |
|
detection
|
This PR improve performance of prior_box op about 1.25x faster on CPU. (#15909)
|
6 years ago |
|
distributed
|
Profiler refine and add CUDA runtime api tracer (#15301)
|
6 years ago |
|
distributed_ops
|
Merge pull request #15536 from jacquesqiao/fix-prefetch-one-parameter
|
6 years ago |
|
elementwise
|
- MKL-DNN pooling updated to set_prim_desc
|
6 years ago |
|
fused
|
add vbroadcast jitkernel refer code and use it
|
6 years ago |
|
jit
|
refine vbroadcast jitcode
|
6 years ago |
|
math
|
Fix error in CUDA kernel of beam_search. (#15957)
|
6 years ago |
|
metrics
|
Fp16 training (#14992)
|
6 years ago |
|
mkldnn
|
Optimize Quantize Op with primitive reuse. (#15929)
|
6 years ago |
|
nccl
|
Revert the changes of VLOG
|
6 years ago |
|
ngraph
|
fix cpplint test=develop (#16028)
|
6 years ago |
|
optimizers
|
enable sgd jitkernel refer code and test
|
6 years ago |
|
reader
|
Revert "Add Event for TensorCopy" (#16022)
|
6 years ago |
|
reduce_ops
|
test=develop
|
6 years ago |
|
sequence_ops
|
refine infershape of sequence_enumerate, hash and fuse_emb_seq_pool
|
6 years ago |
|
tensorrt
|
delete the usage of the const_cast
|
6 years ago |
|
CMakeLists.txt
|
Merge pull request #15609 from xuezhong/add_sample_logits_op
|
6 years ago |
|
activation_cudnn.cu.cc
|
polish cudnn related code and fix bug. (#15164)
|
6 years ago |
|
activation_cudnn_op.cu.cc
|
polish cudnn related code and fix bug. (#15164)
|
6 years ago |
|
activation_op.cc
|
polish cudnn related code and fix bug. (#15164)
|
6 years ago |
|
activation_op.cu
|
Add fp16 backward support (#14202)
|
6 years ago |
|
activation_op.h
|
polish cudnn related code and fix bug. (#15164)
|
6 years ago |
|
add_position_encoding_op.cc
|
position encoding && log loss
|
6 years ago |
|
add_position_encoding_op.h
|
Exhaustive search for cuDNN conv. (#14286)
|
6 years ago |
|
affine_channel_op.cc
|
Add affine channel op to speed and save memory for faster-rcnn model. (#13919)
|
7 years ago |
|
affine_channel_op.cu
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
affine_grid_cudnn_op.cu.cc
|
Add affine grid generator op (#12238)
|
6 years ago |
|
affine_grid_op.cc
|
polish code
|
6 years ago |
|
affine_grid_op.h
|
Fix build error of affine grid op in mac os. (#14237)
|
6 years ago |
|
alloc_continuous_space_op.cc
|
Add alloc_continuous_space_op (#15900)
|
6 years ago |
|
arg_max_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
arg_max_op.cu
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
arg_min_max_op_base.h
|
…
|
|
|
arg_min_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
arg_min_op.cu
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
argsort_op.cc
|
…
|
|
|
argsort_op.cu
|
…
|
|
|
argsort_op.h
|
…
|
|
|
array_operator.h
|
Revert the changes of VLOG
|
6 years ago |
|
array_to_lod_tensor_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
assign_op.cc
|
…
|
|
|
assign_value_op.cc
|
…
|
|
|
assign_value_op.cu.cc
|
…
|
|
|
assign_value_op.h
|
…
|
|
|
attention_lstm_op.cc
|
fix warnings (#15790)
|
6 years ago |
|
attention_lstm_op.h
|
…
|
|
|
average_accumulates_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
average_accumulates_op.cu
|
…
|
|
|
average_accumulates_op.h
|
…
|
|
|
batch_norm_op.cc
|
Merge remote-tracking branch 'origin/develop' into feature/ir_inplace_pass
|
6 years ago |
|
batch_norm_op.cu
|
Enable BatchNorm to use global mean and variane during training (#14630)
|
6 years ago |
|
batch_norm_op.h
|
…
|
|
|
batch_size_like.h
|
…
|
|
|
beam_search_decode_op.cc
|
add per kernel config and remove const_cast.
|
6 years ago |
|
beam_search_decode_op.h
|
Change *(smart_ptr.get()) -> *smart_ptr
|
6 years ago |
|
beam_search_decode_op_test.cc
|
…
|
|
|
beam_search_op.cc
|
Return parent_idx in beam_search op (#15520)
|
6 years ago |
|
beam_search_op.cu.cc
|
Add the CUDA kernel for beam_search op (#15020)
|
6 years ago |
|
beam_search_op.h
|
Return parent_idx in beam_search op (#15520)
|
6 years ago |
|
bilinear_tensor_product_op.cc
|
…
|
|
|
bilinear_tensor_product_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
bilinear_tensor_product_op.h
|
Optimize bilinear tensor product op (#14485)
|
6 years ago |
|
bpr_loss_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
bpr_loss_op.h
|
Add the CUDA kernel for beam_search op (#15020)
|
6 years ago |
|
cast_op.cc
|
…
|
|
|
cast_op.cu
|
…
|
|
|
cast_op.h
|
Revert "cherry picked windows patches."
|
6 years ago |
|
chunk_eval_op.cc
|
…
|
|
|
chunk_eval_op.h
|
…
|
|
|
clip_by_norm_op.cc
|
…
|
|
|
clip_by_norm_op.cu
|
…
|
|
|
clip_by_norm_op.h
|
rewrite variable type
|
6 years ago |
|
clip_op.cc
|
…
|
|
|
clip_op.cu
|
…
|
|
|
clip_op.h
|
…
|
|
|
concat_op.cc
|
Include MKL-DNN header to concat op only when flag is set
|
6 years ago |
|
concat_op.cu.cc
|
…
|
|
|
concat_op.h
|
Refine Split op (#13967)
|
7 years ago |
|
conv_cudnn_op.cu.cc
|
add per kernel config and remove const_cast.
|
6 years ago |
|
conv_cudnn_op_cache.h
|
add per kernel config and remove const_cast.
|
6 years ago |
|
conv_fusion_op.cc
|
Inception fusion operator. (#14968)
|
6 years ago |
|
conv_fusion_op.cu.cc
|
polish
|
6 years ago |
|
conv_op.cc
|
Enable function coverage for U8/S8 ConvMKLDNNOpKernel
|
6 years ago |
|
conv_op.cu.cc
|
…
|
|
|
conv_op.h
|
Memory optimization of depthwise conv op and group norm op (#15313)
|
6 years ago |
|
conv_shift_op.cc
|
…
|
|
|
conv_shift_op.cu
|
…
|
|
|
conv_shift_op.h
|
…
|
|
|
conv_transpose_cudnn_op.cu.cc
|
Revert conv transpose cudnn (#15514)
|
6 years ago |
|
conv_transpose_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
conv_transpose_op.cu.cc
|
…
|
|
|
conv_transpose_op.h
|
…
|
|
|
cos_sim_op.cc
|
…
|
|
|
cos_sim_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
cos_sim_op.h
|
…
|
|
|
crf_decoding_op.cc
|
fix warnings (#15790)
|
6 years ago |
|
crf_decoding_op.h
|
fix enum style
|
6 years ago |
|
crop_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
crop_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
crop_op.h
|
rewrite ddim
|
6 years ago |
|
cross_entropy_op.cc
|
loosly check in the InferShape of cross_entropy_op. (#15863)
|
6 years ago |
|
cross_entropy_op.cu
|
Add fp16 backward support (#14202)
|
6 years ago |
|
cross_entropy_op.h
|
…
|
|
|
ctc_align_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
ctc_align_op.cu
|
…
|
|
|
ctc_align_op.h
|
…
|
|
|
cudnn_lstm_op.cc
|
code refine
|
6 years ago |
|
cudnn_lstm_op.cu.cc
|
merge develop
|
6 years ago |
|
cudnn_rnn_cache.h
|
rewrite variable type
|
6 years ago |
|
cum_op.h
|
fix test issues on windows
|
6 years ago |
|
cumsum_op.cc
|
…
|
|
|
cumsum_op.cu
|
…
|
|
|
data_norm_op.cc
|
remove mkldnn & fix commit
|
6 years ago |
|
data_norm_op.h
|
data_norm
|
6 years ago |
|
delete_var_op.cc
|
[1.1] Load vars on PSERVER (#14037)
|
6 years ago |
|
dequantize_op.cc
|
Fix comments misunderstanding
|
6 years ago |
|
dequantize_op.h
|
Add Dequantize OP
|
6 years ago |
|
detection_map_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
detection_map_op.h
|
…
|
|
|
dropout_op.cc
|
add ShareLoD for dropout_grad (#14616)
|
6 years ago |
|
dropout_op.cu
|
Some improvements to support bert mixed precision training (#15585)
|
6 years ago |
|
dropout_op.h
|
modify dropout att; test=develop
|
7 years ago |
|
dropout_op_test.cc
|
minor fix
|
6 years ago |
|
edit_distance_op.cc
|
…
|
|
|
edit_distance_op.cu
|
…
|
|
|
edit_distance_op.h
|
…
|
|
|
expand_op.cc
|
support multiple var types for expand op, test=develop
|
6 years ago |
|
expand_op.cu
|
support multiple var types for expand op, test=develop
|
6 years ago |
|
expand_op.h
|
rewrite ddim
|
6 years ago |
|
fake_dequantize_op.cc
|
…
|
|
|
fake_dequantize_op.cu
|
…
|
|
|
fake_dequantize_op.h
|
…
|
|
|
fake_quantize_op.cc
|
Fix bug in fake_quantize_op and add more unit testing (#15912)
|
6 years ago |
|
fake_quantize_op.cu
|
…
|
|
|
fake_quantize_op.h
|
…
|
|
|
fc_op.cc
|
fix warnings (#15790)
|
6 years ago |
|
fc_op.h
|
…
|
|
|
fill_constant_batch_size_like_op.cc
|
…
|
|
|
fill_constant_batch_size_like_op.cu.cc
|
…
|
|
|
fill_constant_batch_size_like_op.h
|
…
|
|
|
fill_constant_op.cc
|
register float16
|
6 years ago |
|
fill_constant_op.cu.cc
|
register float16
|
6 years ago |
|
fill_constant_op.h
|
make fill_constant kernel-based
|
6 years ago |
|
fill_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
fill_zeros_like_op.cc
|
…
|
|
|
fill_zeros_like_op.cu.cc
|
Fp16 training (#14992)
|
6 years ago |
|
fill_zeros_like_op.h
|
…
|
|
|
flatten_op.cc
|
squash commits. test=develop
|
6 years ago |
|
gather.cu.h
|
Fix gather & stack op (#14355)
|
6 years ago |
|
gather.h
|
Fix gather & stack op (#14355)
|
6 years ago |
|
gather_op.cc
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
gather_op.cu
|
Some improvements to support bert mixed precision training (#15585)
|
6 years ago |
|
gather_op.h
|
Return parent_idx in beam_search op (#15520)
|
6 years ago |
|
gather_test.cc
|
…
|
|
|
gaussian_random_batch_size_like_op.cc
|
…
|
|
|
gaussian_random_op.cc
|
clean code test=develop
|
6 years ago |
|
gaussian_random_op.cu
|
…
|
|
|
get_tensor_from_selected_rows_op.cc
|
Fix clip.py (#14718)
|
6 years ago |
|
grid_sampler_cudnn_op.cu.cc
|
fix some inappropriate expressions in api doc for grid_sampler. test=develop
|
7 years ago |
|
grid_sampler_op.cc
|
fix grid_sampler PADDLE_ENFORCE error. test=develop (#15542)
|
6 years ago |
|
grid_sampler_op.h
|
code style fix
|
6 years ago |
|
group_norm_op.cc
|
inplace group_norm (#15754)
|
6 years ago |
|
group_norm_op.cu
|
fix pr 15313
|
6 years ago |
|
group_norm_op.h
|
Memory optimization of depthwise conv op and group norm op (#15313)
|
6 years ago |
|
gru_op.cc
|
update gru op forward kernel
|
6 years ago |
|
gru_op.cu.cc
|
update gru op forward kernel
|
6 years ago |
|
gru_op.h
|
update gru op forward kernel
|
6 years ago |
|
gru_unit_op.cc
|
change interface and api spec for dynamic_gru test=develop
|
6 years ago |
|
gru_unit_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
gru_unit_op.h
|
complete gru_unite_op and test
|
6 years ago |
|
hash_op.cc
|
refine infershape of sequence_enumerate, hash and fuse_emb_seq_pool
|
6 years ago |
|
hash_op.h
|
refine infershape of sequence_enumerate, hash and fuse_emb_seq_pool
|
6 years ago |
|
hierarchical_sigmoid_op.cc
|
merge reyoung optimization
|
6 years ago |
|
hierarchical_sigmoid_op.h
|
test=develop, fix debug mode unitest, hsigmoid (#15574)
|
6 years ago |
|
hinge_loss_op.cc
|
rewrite ddim
|
6 years ago |
|
hinge_loss_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
hinge_loss_op.h
|
…
|
|
|
huber_loss_op.cc
|
Add python interface for huber regression loss
|
6 years ago |
|
huber_loss_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
huber_loss_op.h
|
fix the huber loss compile issue on windows test=develop
|
6 years ago |
|
im2sequence_op.cc
|
…
|
|
|
im2sequence_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
im2sequence_op.h
|
…
|
|
|
increment_op.cc
|
…
|
|
|
increment_op.cu
|
…
|
|
|
increment_op.h
|
…
|
|
|
interpolate_op.cc
|
refine image_resize annotation (#15976)
|
6 years ago |
|
interpolate_op.cu
|
test=develop
|
6 years ago |
|
interpolate_op.h
|
test=develop
|
6 years ago |
|
is_empty_op.cc
|
Rewrite is_empty op to avoid unnecessary data transform. (#15509)
|
6 years ago |
|
is_empty_op.cu.cc
|
Rewrite is_empty op to avoid unnecessary data transform. (#15509)
|
6 years ago |
|
is_empty_op.h
|
Rewrite is_empty op to avoid unnecessary data transform. (#15509)
|
6 years ago |
|
isfinite_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
isfinite_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
isfinite_op.h
|
…
|
|
|
l1_norm_op.cc
|
…
|
|
|
l1_norm_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
l1_norm_op.h
|
…
|
|
|
label_smooth_op.cc
|
test=develop
|
6 years ago |
|
label_smooth_op.cu
|
…
|
|
|
label_smooth_op.h
|
…
|
|
|
layer_norm_op.cc
|
fix warnings (#15790)
|
6 years ago |
|
layer_norm_op.cu
|
…
|
|
|
layer_norm_op.h
|
fix enum style
|
6 years ago |
|
linear_chain_crf_op.cc
|
fix warnings (#15790)
|
6 years ago |
|
linear_chain_crf_op.cu
|
…
|
|
|
linear_chain_crf_op.h
|
…
|
|
|
load_combine_op.cc
|
More restrict check load_combine_op. (#15479)
|
6 years ago |
|
load_op.cc
|
fix save and load ops on windows test=develop
|
6 years ago |
|
lod_array_length_op.cc
|
…
|
|
|
lod_rank_table_op.cc
|
Revert the changes of VLOG
|
6 years ago |
|
lod_reset_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
lod_reset_op.cu
|
…
|
|
|
lod_reset_op.h
|
…
|
|
|
lod_tensor_to_array_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
log_loss_op.cc
|
rewrite ddim
|
6 years ago |
|
log_loss_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
log_loss_op.h
|
…
|
|
|
lookup_sparse_table_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
lookup_table_op.cc
|
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into refactor-prefetch
|
6 years ago |
|
lookup_table_op.cu
|
Some improvements to support bert mixed precision training (#15585)
|
6 years ago |
|
lookup_table_op.h
|
fix problem test=develop
|
6 years ago |
|
lrn_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
lrn_op.cu
|
…
|
|
|
lrn_op.h
|
refine lrn_op cpu forward and speedup
|
6 years ago |
|
lstm_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
lstm_op.cu.cc
|
…
|
|
|
lstm_op.h
|
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into fix_bug_for_lstmp
|
6 years ago |
|
lstm_unit_op.cc
|
…
|
|
|
lstm_unit_op.cu
|
…
|
|
|
lstm_unit_op.h
|
Revert "cherry picked windows patches."
|
6 years ago |
|
lstmp_op.cc
|
add cell clip and proj clip, fix bug for h0
|
6 years ago |
|
lstmp_op.cu
|
…
|
|
|
lstmp_op.h
|
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into fix_bug_for_lstmp
|
6 years ago |
|
margin_rank_loss_op.cc
|
…
|
|
|
margin_rank_loss_op.cu
|
…
|
|
|
margin_rank_loss_op.h
|
…
|
|
|
matmul_op.cc
|
…
|
|
|
max_sequence_len_op.cc
|
…
|
|
|
maxout_op.cc
|
…
|
|
|
maxout_op.cu.cc
|
…
|
|
|
maxout_op.h
|
…
|
|
|
mean_iou_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
mean_iou_op.cu
|
[Feature] Add Temporary Allocator (#14875)
|
6 years ago |
|
mean_iou_op.h
|
…
|
|
|
mean_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
mean_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
mean_op.h
|
Add fp16 backward support (#14202)
|
6 years ago |
|
merge_lod_tensor_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
merge_selected_rows_op.cc
|
Refine merge_selected_rows Doc (#14748)
|
6 years ago |
|
merge_selected_rows_op.cu.cc
|
Fix clip.py (#14718)
|
6 years ago |
|
merge_selected_rows_op.h
|
Fix clip.py (#14718)
|
6 years ago |
|
minus_op.cc
|
…
|
|
|
minus_op.cu
|
…
|
|
|
minus_op.h
|
…
|
|
|
modified_huber_loss_op.cc
|
rewrite ddim
|
6 years ago |
|
modified_huber_loss_op.cu
|
…
|
|
|
modified_huber_loss_op.h
|
…
|
|
|
mul_op.cc
|
merge develop
|
6 years ago |
|
mul_op.cu.cc
|
Add fp16 backward support (#14202)
|
6 years ago |
|
mul_op.h
|
…
|
|
|
multiplex_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
multiplex_op.cu
|
…
|
|
|
multiplex_op.h
|
…
|
|
|
nce_op.cc
|
Merge pull request #14893 from JiabinYang/feature/add_prefech_hs
|
6 years ago |
|
nce_op.h
|
nce add check sample lables, test=develop (#15463)
|
6 years ago |
|
norm_op.cc
|
…
|
|
|
norm_op.cu
|
…
|
|
|
norm_op.h
|
test=develop
|
6 years ago |
|
one_hot_op.cc
|
…
|
|
|
one_hot_op.cu
|
…
|
|
|
one_hot_op.h
|
…
|
|
|
pad2d_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
pad2d_op.cu
|
Make pad2d support for variable paddings. (#14667)
|
6 years ago |
|
pad_constant_like_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
pad_constant_like_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
pad_constant_like_op.h
|
…
|
|
|
pad_op.cc
|
…
|
|
|
pad_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
pad_op.h
|
…
|
|
|
pool_cudnn_op.cu.cc
|
Add fp16 backward support (#14202)
|
6 years ago |
|
pool_op.cc
|
use kernel size in global_pooling. test=develop
|
6 years ago |
|
pool_op.cu.cc
|
…
|
|
|
pool_op.h
|
add adaptive pool 2d & 3d. test=develop
|
6 years ago |
|
pool_with_index_op.cc
|
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into feature/tensor_type
|
6 years ago |
|
pool_with_index_op.cu.cc
|
…
|
|
|
pool_with_index_op.h
|
add adaptive pool 2d & 3d. test=develop
|
6 years ago |
|
positive_negative_pair_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
positive_negative_pair_op.h
|
update CMakeLists.txt
|
6 years ago |
|
prelu_op.cc
|
clean
|
6 years ago |
|
prelu_op.cu
|
add prelu gpu inference
|
6 years ago |
|
prelu_op.h
|
…
|
|
|
print_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
psroi_pool_op.cc
|
Fix merge bug
|
6 years ago |
|
psroi_pool_op.cu
|
this is for psroi_pool op, test=develop (#14796)
|
6 years ago |
|
psroi_pool_op.h
|
rewrite ddim
|
6 years ago |
|
py_func_op.cc
|
try fix py2
|
6 years ago |
|
py_func_op.h
|
try fix py2
|
6 years ago |
|
quantize_op.cc
|
Add Quantize OP
|
6 years ago |
|
quantize_op.h
|
Add Quantize OP
|
6 years ago |
|
random_crop_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
random_crop_op.cu
|
…
|
|
|
random_crop_op.h
|
fix security issue 27, 38 test=develop
|
6 years ago |
|
rank_loss_op.cc
|
…
|
|
|
rank_loss_op.cu
|
…
|
|
|
rank_loss_op.h
|
…
|
|
|
recurrent_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
reorder_lod_tensor_by_rank_op.cc
|
refine tensor_array_write_read (#14643)
|
6 years ago |
|
reshape_op.cc
|
Merge remote-tracking branch 'origin/develop' into feature/ir_inplace_pass
|
6 years ago |
|
reverse_op.cc
|
…
|
|
|
reverse_op.cu
|
…
|
|
|
reverse_op.h
|
…
|
|
|
rnn_memory_helper_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
roi_align_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
roi_align_op.cu
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
roi_align_op.h
|
test=develop
|
7 years ago |
|
roi_pool_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
roi_pool_op.cu
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
roi_pool_op.h
|
…
|
|
|
row_conv_op.cc
|
Fix row_conv doc
|
6 years ago |
|
row_conv_op.cu
|
…
|
|
|
row_conv_op.h
|
…
|
|
|
sample_logits_op.cc
|
remove non-ascii charactor
|
6 years ago |
|
sample_logits_op.cu
|
refine code
|
6 years ago |
|
sample_logits_op.h
|
refine code
|
6 years ago |
|
sampling_id_op.cc
|
…
|
|
|
sampling_id_op.cu
|
…
|
|
|
sampling_id_op.h
|
…
|
|
|
save_combine_op.cc
|
fix save and load ops on windows test=develop
|
6 years ago |
|
save_load_combine_op_test.cc
|
…
|
|
|
save_load_op_test.cc
|
…
|
|
|
save_op.cc
|
fix save and load ops on windows test=develop
|
6 years ago |
|
scale_op.cc
|
squash commits. test=develop
|
6 years ago |
|
scale_op.cu
|
Add fp16 backward support (#14202)
|
6 years ago |
|
scale_op.h
|
Fix input<tensor> (#14208)
|
6 years ago |
|
scatter.cu.h
|
Fix gather & stack op (#14355)
|
6 years ago |
|
scatter.h
|
Fix gather & stack op (#14355)
|
6 years ago |
|
scatter_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
scatter_op.cu
|
…
|
|
|
scatter_op.h
|
…
|
|
|
scatter_test.cc
|
…
|
|
|
selu_op.cc
|
Add selu (#14415)
|
6 years ago |
|
selu_op.cu
|
Add selu (#14415)
|
6 years ago |
|
selu_op.h
|
Add selu (#14415)
|
6 years ago |
|
shape_op.cc
|
fix shape api doc
|
6 years ago |
|
shape_op.cu
|
…
|
|
|
shape_op.h
|
…
|
|
|
shrink_rnn_memory_op.cc
|
refine tensor_array_write_read (#14643)
|
6 years ago |
|
shuffle_channel_op.cc
|
rewrite the comments, test=develop
|
6 years ago |
|
shuffle_channel_op.cu
|
update API.spec, test=develop
|
6 years ago |
|
shuffle_channel_op.h
|
Fix, test=develop
|
6 years ago |
|
sigmoid_cross_entropy_with_logits_op.cc
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
sigmoid_cross_entropy_with_logits_op.cu
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
sigmoid_cross_entropy_with_logits_op.h
|
Add generate_mask_labels_op to support Mask-RCNN and refine some code. (#15371)
|
6 years ago |
|
sign_op.cc
|
"fix op. test=develop"
|
7 years ago |
|
sign_op.cu
|
"fix op. test=develop"
|
7 years ago |
|
sign_op.h
|
…
|
|
|
similarity_focus_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
similarity_focus_op.h
|
add similarity_focus op
|
7 years ago |
|
slice_op.cc
|
add lod for slice op, test=develop
|
6 years ago |
|
slice_op.cu
|
…
|
|
|
slice_op.h
|
…
|
|
|
smooth_l1_loss_op.cc
|
…
|
|
|
smooth_l1_loss_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
smooth_l1_loss_op.h
|
…
|
|
|
softmax_cudnn_op.cu.cc
|
Add fp16 backward support (#14202)
|
6 years ago |
|
softmax_op.cc
|
squash commits. test=develop
|
6 years ago |
|
softmax_op.cu.cc
|
Add fp16 backward support (#14202)
|
6 years ago |
|
softmax_op.h
|
- ASUM MKL integration
|
6 years ago |
|
softmax_with_cross_entropy_op.cc
|
change default option related to softmax, test=develop
|
6 years ago |
|
softmax_with_cross_entropy_op.cu
|
[Feature] support mix precision training for resnet (#14899)
|
6 years ago |
|
softmax_with_cross_entropy_op.h
|
- Softmax for Inference is enabled when ON_INFER is set
|
6 years ago |
|
space_to_depth_op.cc
|
Windows/online (#14474)
|
6 years ago |
|
space_to_depth_op.cu
|
test=develop
|
7 years ago |
|
space_to_depth_op.h
|
test=develop
|
7 years ago |
|
spectral_norm_op.cc
|
fix doc statement. test=develop
|
6 years ago |
|
spectral_norm_op.cu
|
fix spectral_norm doc. test=develop
|
6 years ago |
|
spectral_norm_op.h
|
fix format. test=develop
|
6 years ago |
|
split_lod_tensor_op.cc
|
fix bug in if-else op, test=develop
|
6 years ago |
|
split_op.cc
|
Refine Split op (#13967)
|
7 years ago |
|
split_op.cu.cc
|
…
|
|
|
split_op.h
|
Refine Split op (#13967)
|
7 years ago |
|
split_selected_rows_op.cc
|
update height_sections to int64_t
|
7 years ago |
|
split_selected_rows_op.cu
|
…
|
|
|
split_selected_rows_op.h
|
split selected rows op should always init output selected rows
|
6 years ago |
|
spp_op.cc
|
…
|
|
|
spp_op.cu.cc
|
…
|
|
|
spp_op.h
|
add adaptive pool 2d & 3d. test=develop
|
6 years ago |
|
squared_l2_distance_op.cc
|
…
|
|
|
squared_l2_distance_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
squared_l2_distance_op.h
|
…
|
|
|
squared_l2_norm_op.cc
|
…
|
|
|
squared_l2_norm_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
squared_l2_norm_op.h
|
…
|
|
|
squeeze_op.cc
|
…
|
|
|
stack_op.cc
|
Fix gather & stack op (#14355)
|
6 years ago |
|
stack_op.cu
|
Some improvements to support bert mixed precision training (#15585)
|
6 years ago |
|
stack_op.h
|
Add the macro for NVCC (test=develop)
|
6 years ago |
|
strided_memcpy.h
|
rewrite ddim
|
6 years ago |
|
strided_memcpy_test.cc
|
…
|
|
|
sum_op.cc
|
fix sum_op selected rows test=develop
|
6 years ago |
|
sum_op.cu
|
Fix Eigen macro when using GPU
|
6 years ago |
|
sum_op.h
|
rewrite variable type
|
6 years ago |
|
teacher_student_sigmoid_loss_op.cc
|
remove mkl & fix commit
|
6 years ago |
|
teacher_student_sigmoid_loss_op.h
|
remove some comments & refine doc & put template class in .h
|
6 years ago |
|
tensor_array_to_tensor_op.cc
|
Use sub scope in tensor_array_to_tensor op. (#14524)
|
6 years ago |
|
top_k_op.cc
|
Make topk op support variable k. (#15044)
|
6 years ago |
|
top_k_op.cu
|
Make topk op support variable k. (#15044)
|
6 years ago |
|
top_k_op.h
|
Make topk op support variable k. (#15044)
|
6 years ago |
|
transpose_op.cc
|
- Added transpose/transpose2 MKLDNN grad ops
|
6 years ago |
|
transpose_op.cu.cc
|
Some improvements to support bert mixed precision training (#15585)
|
6 years ago |
|
transpose_op.h
|
…
|
|
|
tree_conv_op.cc
|
Tree conv op (#15217)
|
6 years ago |
|
tree_conv_op.cu
|
Tree conv op (#15217)
|
6 years ago |
|
tree_conv_op.h
|
Tree conv op (#15217)
|
6 years ago |
|
truncated_gaussian_random_op.cc
|
…
|
|
|
truncated_gaussian_random_op.cu
|
…
|
|
|
uniform_random_batch_size_like_op.cc
|
…
|
|
|
uniform_random_op.cc
|
shape type to int64_t, test=develop
|
7 years ago |
|
uniform_random_op.cu
|
fix shape type in uniform_random_op.cu
|
7 years ago |
|
unpool_op.cc
|
Change tensor uses proto::VarType::type
|
6 years ago |
|
unpool_op.cu.cc
|
…
|
|
|
unpool_op.h
|
…
|
|
|
unsqueeze_op.cc
|
…
|
|
|
unstack_op.cc
|
…
|
|
|
unstack_op.h
|
…
|
|
|
warpctc_cudnn_op.cu.cc
|
Revert conv transpose cudnn (#15514)
|
6 years ago |
|
warpctc_op.cc
|
polish code
|
6 years ago |
|
warpctc_op.cu.cc
|
…
|
|
|
warpctc_op.h
|
…
|
|
dengkaipeng