提交 · 2916caa2c488df6a11fa74c825eda95bb86aa2cb · BaiXuePrincess / Paddle

30 8月, 2019 1 次提交
- H
  Change ugly PADDLE_ENFORCE_EQ in recurrent_op.cc (#19470) · 2916caa2
  由 Huihuang Zheng 提交于 8月 30, 2019
```
test=develop
```
  2916caa2
27 8月, 2019 1 次提交
- H
  
  Delete useless ex-scope in recurrent op (#19426) · 56dd7653
  由 Huihuang Zheng 提交于 8月 27, 2019
  
  56dd7653
26 8月, 2019 1 次提交
- H
  
  Change TensorCopy in recurrent_op to ShareDataWith (#19319) · 12d29f4d
  由 Huihuang Zheng 提交于 8月 26, 2019
  
  12d29f4d
19 7月, 2019 1 次提交

Support memory eager deletion on recurrent OP (#17710) · 89bc3fd8

由 Huihuang Zheng 提交于 7月 19, 2019

Test PaddingRNN on V100 GPU device.

Test configuration: large model, padding mode (which is the mode using recurrentOp), one GPU.

GPU memory (MiB): 6414 (this PR) vs 6837 (without this PR)
Speed (steps/s): 10.28 (this PR) vs 9.89 (without this PR)

89bc3fd8

30 5月, 2019 1 次提交
- Y
  Optimize recurrent_op using Prepare and RunPreparedContext, avoiding create... · 2704479b
  由 Yiqun Liu 提交于 5月 30, 2019
```
Optimize recurrent_op using Prepare and RunPreparedContext, avoiding create operators in every iter. (#17689)

test=develop
```
  2704479b
19 5月, 2019 1 次提交
- Z
  
  fix recurrent fwd bug when no backward and scope clear (#17460) · 3d4e8268
  由 Zeng Jinle 提交于 5月 19, 2019
  
  3d4e8268
16 5月, 2019 1 次提交
- Z
  
  fix recurrent_op,test=develop (#17433) · 712bfb17
  由 Zeng Jinle 提交于 5月 16, 2019
  
  712bfb17
12 4月, 2019 1 次提交
- C
  Refine StaticRnn (#16707) · c62674f4
  由 chengduo 提交于 4月 12, 2019
```
* enable recurrent op test=develop
```
  c62674f4
28 3月, 2019 1 次提交
- S
  revert some loop op revision · 4c8254e3
  由 sneaxiy 提交于 3月 27, 2019
```
test=develop
```
  4c8254e3
27 3月, 2019 1 次提交
- S
  fix grad desc maker · 63651c19
  由 sneaxiy 提交于 3月 27, 2019
```
test=develop
```
  63651c19
11 3月, 2019 1 次提交
- S
  disable gc in recurrent_op currently · 732fa00e
  由 sneaxiy 提交于 3月 08, 2019
```
test=develop
```
  732fa00e
06 3月, 2019 2 次提交
- C
  Refine recurrent_op (#16027) · 6fe7478b
  由 chengduo 提交于 3月 06, 2019
```
* refine recurrent_op
test=develop

* remove unnecessary code
test=develop
```
  6fe7478b
- C
  Refine recurrent_op (#16027) · f5a37518
  由 chengduo 提交于 3月 06, 2019
```
* refine recurrent_op
test=develop

* remove unnecessary code
test=develop
```
  f5a37518
12 12月, 2018 1 次提交
- Y
  Change tensor uses proto::VarType::type · 9bd70a1e
  由 Yu Yang 提交于 12月 11, 2018
```
test=develop
```
  9bd70a1e
26 11月, 2018 1 次提交
- M
  Revert the changes of VLOG · 53433d7f
  由 minqiyang 提交于 11月 26, 2018
```
test=develop
```
  53433d7f
08 11月, 2018 1 次提交
- M
  Change the origin VLOG level to 10 times · 0c3227a5
  由 minqiyang 提交于 11月 08, 2018
```
Fix code to support cpplint syntax check

test=develop
```
  0c3227a5
21 6月, 2018 2 次提交
- T
  Revert "Merge pull request #11628 from PaddlePaddle/revert-11102-mozga-intel/Sum_mkldnn_layout" · d5fb8fa7
  由 tensor-tang 提交于 6月 21, 2018
```
This reverts commit 4d8e8ee2, reversing
changes made to d6a9f005.
```
  d5fb8fa7
- T
  
  Revert "MKLDNN layout: Support for sum operator" · 90780e22
  由 tensor-tang 提交于 6月 21, 2018
  
  90780e22
19 6月, 2018 1 次提交
- M
  
  MKLDNN layout: Support for sum operator · 96b4904d
  由 mozga-intel 提交于 6月 12, 2018
  
  96b4904d
08 5月, 2018 1 次提交

Clean OpProtoAndCheckerMaker · 0e78cb69

由 Yu Yang 提交于 5月 08, 2018

Do not use ctor

* Reduce line of codes.
* We can use virtual function for Maker now.
* The implementation does not care what maker holds, it is easier to
refactor later.

0e78cb69

25 4月, 2018 1 次提交
- A
  
  Fix CPPLint errors with op_desc · edd3587e
  由 Abhinav Arora 提交于 4月 24, 2018
  
  edd3587e
15 2月, 2018 1 次提交

Update tensor_util.h (#8422) · cfffb1a3

由 Yi Wang 提交于 2月 14, 2018

* Update tensor_util.h

* Update with moved TensorDesc

* Fix tensur_utils.cu

* Update

* Update

* Update

* Update

* Make tensor_util.cu a symbolic link

cfffb1a3

12 2月, 2018 1 次提交
- Q
  
  Fix the grammar in copyright. (#8403) · 24509f4a
  由 qingqing01 提交于 2月 12, 2018
  
  24509f4a
10 2月, 2018 2 次提交
- Y
  
  Correct #include path · fc374821
  由 Yi Wang 提交于 2月 09, 2018
  
  fc374821
- Y
  
  Move file to fluid/; Edit CMakeLists.txt · 90648f33
  由 Yi Wang 提交于 2月 09, 2018
  
  90648f33
09 2月, 2018 1 次提交
- Y
  
  use op run as wrapper of run_impl; make run_impl as private virtual function · 98c94373
  由 Yang Yang 提交于 2月 09, 2018
  
  98c94373
09 1月, 2018 1 次提交
- Y
  Rename CopyFrom to Copy for tensors (#7292) · ce6dad3b
  由 Yu Yang 提交于 1月 09, 2018
```
* Rename Tensor::CopyFrom to Tensor::Copy

* Fix CI

* Fix compile
```
  ce6dad3b
27 12月, 2017 4 次提交
- Y
  Rename API of DeviceContext (#7055) · 15e8c80e
  由 Yu Yang 提交于 12月 27, 2017
```
* Rename API of DeviceContext

Make them as usual names.

* Rename API of DeviceContext

Make them as usual names.

* Fix compile

* Fix compile

* Fix compile

* Fix compile

* Fix compile
```
  15e8c80e
- Y
  Rename API of DeviceContext · 8b877dd7
  由 Yang Yu 提交于 12月 27, 2017
```
Make them as usual names.
```
  8b877dd7
- Y
  Rename API of DeviceContext · a5e1cf5a
  由 Yang Yu 提交于 12月 27, 2017
```
Make them as usual names.
```
  a5e1cf5a
- Y
  Rename API of DeviceContext · fd2bf550
  由 Yang Yu 提交于 12月 27, 2017
```
Make them as usual names.
```
  fd2bf550
26 12月, 2017 2 次提交
- Y
  Polish `Scope::LocalVarNames` · ef188371
  由 Yang Yu 提交于 12月 26, 2017
```
Cannot get var name recursive since they could be same.
```
  ef188371
- L
  
  unify the indentation of license · 761b3297
  由 Luo Tao 提交于 12月 26, 2017
  
  761b3297
24 12月, 2017 1 次提交

Feature/operator run place (#6783) · 735eba29

由 dzhwinter 提交于 12月 24, 2017

* "change operator interface"

* "move devicepool to device_context"

* "fix operator test"

* "fix op_registry Run interface"

* "net op passed. Need to fix nccl multi-Context"

* "add nccl group function"

* "add nccl group function"

* "fix gpu count exceed 32 error"

* "fix recurrent op, nccl op"

* "change the other operators interface with Place"

* "fix typo"

* "fix pybind"

* "fix device in python side"

* "fix pybind failed"

* "add init for test"

* "fix CI"

735eba29

22 12月, 2017 1 次提交

Enforce drop_empty_grad=false When the input of an op is duplicable. · 0bfa1f7c

由 xuwei06 提交于 12月 01, 2017

For input argument with a list of variables, drop_empty_grad is not allowed because it makes the correspondence bewteen a variable and its gradient ambiguous. Use REGISTER_OP_EX to register the op or call InputGrad(?,false) in GradOpDescMaker.

0bfa1f7c

21 12月, 2017 1 次提交
- Y
  Rename XXDescBind --> XXDesc (#6797) · 09189732
  由 Yu Yang 提交于 12月 21, 2017
```
* Rename XXDescBind --> XXDesc

* Fix Compile
```
  09189732
20 12月, 2017 1 次提交
- Y
  Move framework.proto to proto namespace (#6718) · e445b3ff
  由 Yu Yang 提交于 12月 20, 2017
```
* Move framework.proto to proto namespace

* Fix compile

* Fix compile

* Fix Compile
```
  e445b3ff
19 12月, 2017 2 次提交
- Y
  
  revise typo · aea5ccca
  由 Yang Yang 提交于 12月 19, 2017
  
  aea5ccca
- Y
  
  parallel_do skeleton pass compile · 9d2c77e6
  由 Yang Yang 提交于 12月 19, 2017
  
  9d2c77e6
14 12月, 2017 1 次提交
- F
  
  Unify `step_block` and `block` to `sub_block` · dafd449c
  由 fengjiayi 提交于 12月 14, 2017
  
  dafd449c

BaiXuePrincess / Paddle 与 Fork 源项目一致

BaiXuePrincess / Paddle
与 Fork 源项目一致