提交 · f98ff2ce4eea837eee2304ca1c69ffc9f64bdf80 · PaddlePaddle / Paddle

28 3月, 2023 34 次提交
- H
  
  [Fix Bug] Fix CustomOP cmake copy file error in inference (#52171) · f98ff2ce
  由 HongyuJia 提交于 3月 28, 2023
  
  f98ff2ce
- H
  fix int8 support for full kernel (#52194) · c145fd1e
  由 houj04 提交于 3月 28, 2023
```
* fix int8 support for full kernel

* fix ut.
```
  c145fd1e
- G
  
  [CodeStyle][B004] hasattr replace callable (#52096) · 4118ab89
  由 gouzil 提交于 3月 28, 2023
  
  4118ab89
- N
  
  fix a typo, `sheduler` -> `scheduler` (#52149) · e492ee24
  由 Nyakku Shigure 提交于 3月 28, 2023
  
  e492ee24
- W
  Polish ResNet and Bert prim_cinn test (#52030) · e57051b4
  由 WangZhen 提交于 3月 28, 2023
```
* Polish ResNet and Bert prim_cinn test
```
  e57051b4
- C
  support auto generate for huber_loss (#51951) · 2ba4515e
  由 cyberslack_lee 提交于 3月 28, 2023
```
* fix huber_loss

* fix

* fix ops.yaml add intermediate

* fix

* fix test
```
  2ba4515e
- R
  support auto generate static for one_hot_v2 (#52134) · b6af72eb
  由 RedContritio 提交于 3月 28, 2023
```
* support auto generate static for one_hot_v2

* format
```
  b6af72eb
- W
  
  add autogen code support for margin_cross_entropy (#52130) · 8c8c6d9d
  由 Wang Xin 提交于 3月 28, 2023
  
  8c8c6d9d
- R
  [静态图算子自动生成] support auto generate for log_softmax (#52036) · ad9b88ad
  由 RedContritio 提交于 3月 28, 2023
```
* support auto generate for log_softmax

* add data_type
```
  ad9b88ad
- A
  
  【Hackathon 4th 】add Trapezoid API && add Cumulative_trapezoid API (#51195) · cfe3ff48
  由 Ainavo 提交于 3月 28, 2023
  
  cfe3ff48
- K
  
  [CodeStyle][PLR1701] unify multiple isinstance expressions as one (#52150) · c1838da6
  由 Kim 提交于 3月 28, 2023
  
  c1838da6
- G
  [CodeStyle][B020] rename iterable and it's iterates when they have same name (#52128) · c05feb90
  由 gouzil 提交于 3月 28, 2023
```
* [CodeStyle][B020] rename for

* [CodeStyle][B020] rename for
```
  c05feb90
- H
  
  [API/OP] Support FP16/BF16 in paddle.nonzero API/OP (#51640) · 2e92357b
  由 Haohongxiang 提交于 3月 28, 2023
  
  2e92357b
- W
  [AMP OP&Test] add fp16/bf16 unittest for conv ops (#51787) · ad5536eb
  由 wangxinxin08 提交于 3月 28, 2023
```
* add unittest for conv2d/depthwise_conv2d/conv2d_transpose

* add bf16 for DWConv and ConvTranspose

* fix unitest of conv2d_transpose

* modify DWConv2d op and unittest

* fix unittest of conv2d_transpose_bf16

* modify unittest name according to review

* modify atol of DWConv2D unittest
```
  ad5536eb
- G
  [Test Mv] autograd_test (#52142) · b94ef537
  由 gouzil 提交于 3月 28, 2023
```
* [Test Mv] autograd_test

* [Move Test] rm py_test_modules
```
  b94ef537
- Z
  
  [XPU] fix bug of AnalyseOpFuncType about xpu op : memcpy_d2d of xpu is actually async (#52042) · 93d20c44
  由 ZhouMengLei1999 提交于 3月 28, 2023
  
  93d20c44
- W
  delete old dygraph mkldnn op test (#51953) · 67a105f9
  由 wanghuancoder 提交于 3月 28, 2023
```
* delete old dygraph mkldnn op test
```
  67a105f9
- W
  [AMP OP&Test] Add float16 OpTest for squeeze, unsqueeze (#52018) · 866c2877
  由 Wang Xinyu 提交于 3月 28, 2023
```
* add squeeze, unsqueeze, transpose fp16 unitest

* Update test_transpose_op.py
```
  866c2877
- C
  [Auto Parallel] Add o1 level tune (#52041) · d6011cb6
  由 caozhou 提交于 3月 28, 2023
```
* add tune o1 level

* add unittest
```
  d6011cb6
- Y
  
  [AMP OP&Test]shape op fp/bf16 support (#52184) · 418b983c
  由 YuhangLi 提交于 3月 28, 2023
  
  418b983c
- A
  [Test Mv] move collective/multinode to test dir (#51982) · ceca55c5
  由 Ainavo 提交于 3月 28, 2023
```
* [Test Mv] move collective/multinode to test dir

* add CMakeList.txt to test/collective

* add bash_test_modules

* adjust the order

* recover bash_test_modules

* add_subdirectory(collective)

* resolve conflicts

* resolve conflicts
```
  ceca55c5
- K
  [Test Mv]mv dataset test (#52148) · 431e8dbd
  由 Kim 提交于 3月 28, 2023
```
* mv python/paddle/dataset/tests/ test/dataset/

* use path to get file path

* add dataset

* remove cmake entry

* add all ops

* solve conflicts
```
  431e8dbd
- J
  [AMP] add fp16&bf16 support for flatten op (#52035) · a33a4d01
  由 jiangcheng 提交于 3月 28, 2023
```
* [AMP] add fp16&bf16 support for flatten op

* fix ci bug

* fix inpute should astype self.dtype bug and fix zerodim test name

* remove 0D-tensor bf16 test for window-inference-ci pass

* remove flatten from op_accuracy_white_list
```
  a33a4d01
- Z
  [Move Test] move rpc (#52166) · a34abdb5
  由 Zheng-Bicheng 提交于 3月 28, 2023
```
* update

* update
```
  a34abdb5
- Z
  [Move Test] move contrib (#52168) · 28a76556
  由 Zheng-Bicheng 提交于 3月 28, 2023
```
* update

* update
```
  28a76556
- G
  [CodeStyle][B015] replace pointless comparisons with appropriate statements (#52126) · 3957007c
  由 gouzil 提交于 3月 28, 2023
```
* [CodeStyle][B015] delete unused

* [CodeStyle][B015] add assert
```
  3957007c
- 张
  
  [CodeStyle][PLC0414] remove self-alias and some discussion (#52122) · 888b8b6b
  由张春乔提交于 3月 28, 2023
  
  888b8b6b
- R
  
  [CustomDevice] fix reducer (#52115) · e7c249cb
  由 ronnywang 提交于 3月 28, 2023
  
  e7c249cb
- I
  
  [CodeStyle][C405] Unnecessary <list/tuple> literal - rewrite as a set literal (#51972) · 9fa98349
  由 Infinity_lee 提交于 3月 28, 2023
  
  9fa98349
- Z
  
  [Test Mv] python/paddle/fluid/tests/custom_op/*.py to test/custom_op (#51948) · 7aa7fc49
  由 Zheng-Bicheng 提交于 3月 28, 2023
  
  7aa7fc49
- J
  【Prim】Optimize composite rule by making scalar shape as 1 (#51960) · 45acb717
  由 Jiabin Yang 提交于 3月 28, 2023
```
* optimize composite rule by making scalar shape as []1

* fix shape usage for 0D

* fix rules

* fix 0D error

* fix flatten 0D error

* fix bn eval mode

* fix bn test

* fix flatten
```
  45acb717
- Y
  [Hackathon NO.77] 为 Paddle-TRT 添加 bitwise 算子 (#51971) · 864b50c3
  由 Young-Flash 提交于 3月 28, 2023
```
* add bitwise_not trt converter

* run pre-commit

* modify neg_one_tensor_dims init way

* fix BOOL type support requires TensorRT 8.4

* fix int8 & uint8 type

* improve data type readability

* modify filter logic

* fix coverage CI
```
  864b50c3
- C
  
  Modify the registration information of the interpolate kernel (#52163) · 3b055199
  由 csy0225 提交于 3月 28, 2023
  
  3b055199
- I
  
  [CodeStyle][UP024] Replace aliased errors with OSError (#52024) · 8c888eea
  由 Infinity_lee 提交于 3月 28, 2023
  
  8c888eea
27 3月, 2023 6 次提交

G
[CodeStyle][PLC3002][PLE1205] simplify lambda and add missing placeholder to... · b166581a
由 gouzil 提交于 3月 27, 2023
```
[CodeStyle][PLC3002][PLE1205] simplify lambda and add missing placeholder to logger template (#52133)
```
b166581a

[CodeStyle][C413][C414] Unnecessary <list/reversed> call around... · af6f262d

由 Infinity_lee 提交于 3月 27, 2023

[CodeStyle][C413][C414] Unnecessary <list/reversed> call around sorted(),<list/reversed/set/sorted/tuple> call within <list/set/sorted/tuple>() (#52065)

af6f262d

add prim test for some ops (#51749) · e1674e8b

由 Charles-hit 提交于 3月 27, 2023

* add tanh and cast prim test

* fix tanh test

* fix 0-d test

* add sqrt fp16 prim test

* add public_python_api in prim test

* fix test_squeeze2_op

* add tanh prim test

* add dropout prim test

* [Dy2St]Fix clone for test state problem

* clean code

* modify test_cumsum_op

* modify test_cumsum_op

* fix dropout test

* add dropout in cmake

* fix dropout test

---------
Co-authored-by: NAurelius84 <zhangliujie@baidu.com>

e1674e8b

C

fix eval branch of composite rule of batch_norm (#52154) · 20befdef
由 cyber-pioneer 提交于 3月 27, 2023

20befdef
Y
[PHI]Support register functor kernel into PHI (#51914) · bcea3b89
由 YuanRisheng 提交于 3月 27, 2023
```
* perfect structure kernel registry

* fix ci bugs
```
bcea3b89
A

[NewExe]Adjust ExecutorCache Capacity from 4 into 10 (#52104) · 897fb6ab
由 Aurelius84 提交于 3月 27, 2023

897fb6ab

PaddlePaddle / Paddle 大约 1 年 前同步成功

PaddlePaddle / Paddle
大约 1 年前同步成功