提交 · c4c5f0b8ca79fdb35397f0e96ce1050a4a8961dc · PaddlePaddle / PaddleDetection

10 12月, 2018 6 次提交
- B
  Remove unnecessary MKLDNN reorder (#14799) · 9623b45f
  由 Brian Liu 提交于 12月 10, 2018
```
When data flow from a MKLDNN OP kernel to a non-MKLDNN OP kernel,
data layout transform (via MKLDNN reorder) will occur even when
those two OP kernels share same layout. Add code to remove this
unnecessary reorder.

test=develop
```
  9623b45f
- F
  
  fix 'name', test=develop · 90c7f987
  由 frankwhzhang 提交于 12月 10, 2018
  
  90c7f987
- N
  
  fix comments test=develop · 019e8bbe
  由 nhzlx 提交于 12月 10, 2018
  
  019e8bbe
- F
  
  update API, test=develop · 271c4808
  由 frankwhzhang 提交于 12月 10, 2018
  
  271c4808
- F
  
  fix label_pos ,add test_layers.py, test=develop · c9a65382
  由 frankwhzhang 提交于 12月 10, 2018
  
  c9a65382
- S
  add cuda cudnn version check · 66182abd
  由 sneaxiy 提交于 12月 10, 2018
```
test=develop
```
  66182abd
09 12月, 2018 4 次提交
- F
  
  fix code style, test=develop · a672b291
  由 frankwhzhang 提交于 12月 09, 2018
  
  a672b291
- F
  
  fix style bug, test=develop · ea95f9c3
  由 frankwhzhang 提交于 12月 09, 2018
  
  ea95f9c3
- F
  
  fix nn.py&API.spec, test=develop · 68c20258
  由 frankwhzhang 提交于 12月 09, 2018
  
  68c20258
- X
  Revert "Merge pull request #14798 from... · 748549b2
  由 Xin Pan 提交于 12月 09, 2018
```
Revert "Merge pull request #14798 from PaddlePaddle/revert-14786-revert-14782-revert-14398-imperative"

This reverts commit b1d3a1c8, reversing
changes made to f1fb64b1.
```
  748549b2
08 12月, 2018 1 次提交

One possible solution to add flexibility for mkldnn placement pass (#14768) · 943ad478

由 bingyanghuang 提交于 12月 08, 2018

* Choose to turn on use_mkldnn attribute v1

* Fix mkldnn_op empty bug

* format change test=develop

* fix ci test=develop

* fix ci test and add test in dam test=develop

* add example to dam compare test test=develop

* review changes test=develop

943ad478

07 12月, 2018 6 次提交
- X
  
  Revert "Revert "Revert "Imperative""" · c049fa7c
  由 Xin Pan 提交于 12月 07, 2018
  
  c049fa7c
- G
  
  Add reduce sparse tensor feature. (#14757) · f1fb64b1
  由 gongweibao 提交于 12月 07, 2018
  
  f1fb64b1
- F
  
  update bpr_loss op code, test=develop · 97de98cd
  由 frankwhzhang 提交于 12月 07, 2018
  
  97de98cd
- Y
  Clean Code · 240d974a
  由 Yihua Xu 提交于 12月 07, 2018
```
test=develop
```
  240d974a
- X
  
  Revert "Revert "Imperative"" · 2538ef64
  由 Xin Pan 提交于 12月 07, 2018
  
  2538ef64
- Y
  Clean Code · 155328a4
  由 Yihua Xu 提交于 12月 07, 2018
```
test=develop
```
  155328a4
06 12月, 2018 12 次提交
- X
  
  Revert "Imperative" · 6217f42a
  由 Xin Pan 提交于 12月 06, 2018
  
  6217f42a
- T
  update with comments · 743cb840
  由 Tao Luo 提交于 12月 06, 2018
```
test=develop
```
  743cb840
- W
  AsyncExecutor bugfix: Tensor to LoDTensor · 5a2cd450
  由 wangguibao 提交于 12月 06, 2018
```
test=develop
```
  5a2cd450
- W
  
  AsyncExecutor bugfix: Tensor change to LoDTensor · 5f98d800
  由 wangguibao 提交于 12月 06, 2018
  
  5f98d800
- F
  
  add tool to visualize inference model (#14621) · f6a877bc
  由 flame 提交于 12月 06, 2018
  
  f6a877bc
- F
  
  update API.spec · 93551a34
  由 frankwhzhang 提交于 12月 06, 2018
  
  93551a34
- T
  clean code · 42359e88
  由 Tao Luo 提交于 12月 06, 2018
```
test=develop
```
  42359e88
- T
  support loading from memory · 405b2486
  由 Tao Luo 提交于 12月 04, 2018
```
test=develop
```
  405b2486
- F
  
  add bpr_loss operator , test=develop · 570d89ec
  由 frankwhzhang 提交于 12月 06, 2018
  
  570d89ec
- Q
  Speed conv_fusion_op for identity activation. (#14744) · 549f165b
  由 qingqing01 提交于 12月 06, 2018
```
* Refine conv_fusion_op for identity activation.
* Fix unit testing.
test=develop
```
  549f165b
- S
  fix thread-safety bug · 0f96c2e8
  由 sneaxiy 提交于 12月 05, 2018
```
test=develop
```
  0f96c2e8
- L
  
  fix the bug for mac build. python -c error. test=develop · 5026741b
  由 lujun 提交于 12月 06, 2018
  
  5026741b
05 12月, 2018 9 次提交
- N
  fix bug of trt pool · 722b0a80
  由 nhzlx 提交于 12月 05, 2018
```
test=develop
```
  722b0a80
- J
  
  fix bug in dist train on hs, test=develop · e05e1d7d
  由 JiabinYang 提交于 12月 05, 2018
  
  e05e1d7d
- T
  remove jit namespace · 4a93db92
  由 tensor-tang 提交于 12月 05, 2018
```
test=develop
```
  4a93db92
- X
  follow comments · 82d68281
  由 Xin Pan 提交于 12月 05, 2018
```
test=develop
```
  82d68281
- S
  fix deallocate bug · 90076522
  由 sneaxiy 提交于 12月 05, 2018
```
test=develop
```
  90076522
- L
  
  fix bug of formate; test=develop · 8b2898e2
  由 liuhongyu 提交于 12月 05, 2018
  
  8b2898e2
- X
  allow customize kernel selection · 41c28d54
  由 Xin Pan 提交于 12月 05, 2018
```
test=develop
```
  41c28d54
- L
  
  fix bugs; test=develop · 8daf67f9
  由 liuhongyu 提交于 12月 05, 2018
  
  8daf67f9
- C
  Fix clip.py (#14718) · 04539d4c
  由 chengduo 提交于 12月 05, 2018
```
* expose square
test=develop

* fix activation
test=develop

* Add square API
test=develop

* add necessary op

* code refine

* fix API.spec
test=develop

* fix unit test
test=develop

* add unit test sparse_grad_clip
test=develop

* fix API.spec
test=develop

* remove mac test for test_gradient_clip
test=develop

* remove selectedrows_mul_tensor
test=develop
```
  04539d4c
04 12月, 2018 2 次提交

[Feature] multi process multi gpu dist training, boost v100 performance by 20% (#14661) · 29d9fb53

由 Wu Yi 提交于 12月 04, 2018

* wip multi process multi gpu dist training

* workable for p2p

* update test=develop

* change back env name test=develop

* fix alloc init

* fix cpu build test=devlop

* fix mac tests test=develop

* refine code

* refine test=develop

29d9fb53

L

add cudnn 5 support; test=develop · 968dd3c0
由 liuhongyu 提交于 12月 04, 2018

968dd3c0

PaddlePaddle / PaddleDetection 大约 1 年 前同步成功

PaddlePaddle / PaddleDetection
大约 1 年前同步成功