提交 · 73e41c893b258628c8faeb8841b0708c9f20ad48 · 机器未来 / Paddle

24 8月, 2022 12 次提交
- S
  Solve the random state serialization (#45327) · 73e41c89
  由 ShenLiang 提交于 8月 24, 2022
```
* fix utest

* fix utest

* fix utest

* fix log

* fix random utest
```
  73e41c89
- L
  make tensor_util contains no cuda code (#45256) · 78916a7a
  由 Leo Chen 提交于 8月 24, 2022
```
* make tensor_util contains no cuda code

* refine isfinite

* revert ut

* move isfinite function to its op

* fix test

* fix compile

* std::isnan is not defined for int type on windows

* fix windows compile

* fix fp16

* fix rocm compile

* revert gradient node
```
  78916a7a
- Y
  
  fix op_teller with_dynamic_shape judge bug (#45384) · 9e0baf6e
  由 Yuanle Liu 提交于 8月 24, 2022
  
  9e0baf6e
- J
  
  fix argsort grad not fill zero (#45371) · c70d79a0
  由 Jiabin Yang 提交于 8月 24, 2022
  
  c70d79a0
- H
  [phi] Transfer merged_momentum yaml to phi (#45359) · 09acc860
  由 HongyuJia 提交于 8月 24, 2022
```
* add legacy_api.yaml

* set merged_momentum inplace only

* support inplace optional<vector<tensor>>

* add dygraph_mode api

* optimize TensorToConstDenseTensorPtr
```
  09acc860
- W
  
  Adapt tensor axis for cumsum (#45372) · 7f49b9ba
  由 WangZhen 提交于 8月 24, 2022
  
  7f49b9ba
- W
  
  conv_eltwiseadd_bn_fuse support fp16 (#45379) · 62b5452d
  由 Wilber 提交于 8月 24, 2022
  
  62b5452d
- M
  Support fp16 of adam operator in xpu environment (#45292) · a012d426
  由 mengqingchun02 提交于 8月 24, 2022
```
* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun
```
  a012d426
- W
  [OpAttr]Adapt tensor minlength for bincount (#45342) · 12917c8c
  由 WangZhen 提交于 8月 24, 2022
```
* Adapt minlength attr for bincount
```
  12917c8c
- W
  trt input tensor issue (#45358) · 55393172
  由 wenbin 提交于 8月 24, 2022
```
* fix

* optimize
```
  55393172
- Z
  
  [MLU]: compile prior box CPU kernel when WITH_MLU is ON for SSD (#45238) · 596a9405
  由 zhaoying9105 提交于 8月 24, 2022
  
  596a9405
- W
  
  fix convert weight failed. (#45346) · 3d514e48
  由 Wilber 提交于 8月 24, 2022
  
  3d514e48
23 8月, 2022 9 次提交
- P
  
  print log while use new exe (#45335) · fac8a260
  由 pangyoki 提交于 8月 23, 2022
  
  fac8a260
- Z
  [AutoParallel] Add Quant Pass (#44877) · 61bc016c
  由 zhaoyingli 提交于 8月 23, 2022
```
* add quant pass
```
  61bc016c
- L
  
  [FleetExecutor] Using program to be the only interface of TaskNode (#43869) · 9ccdb5fa
  由 LiYuRio 提交于 8月 23, 2022
  
  9ccdb5fa
- N
  
  Delete the template parameter BLockSize in Kernel Primitive API (#45220) · 1a0cd447
  由 niuliling123 提交于 8月 23, 2022
  
  1a0cd447
- O
  
  Update scope.h (#45270) · 60e072d3
  由 OccupyMars2025 提交于 8月 23, 2022
  
  60e072d3
- L
  
  Add store_barrier to prevent master exit (#44964) · 1734bc6f
  由 LiYuRio 提交于 8月 23, 2022
  
  1734bc6f
- O
  modify something unimportant when I read source code (#45273) · 5edc96e6
  由 OccupyMars2025 提交于 8月 23, 2022
```
* Update scope.h

* typo

* Update dense_tensor.inl
```
  5edc96e6
- Y
  [Phi]Move distribute_fpn_proposals to PHI (#45212) · 8f8ed7de
  由 YuanRisheng 提交于 8月 23, 2022
```
* move distribute_fpn_proposals

* fix some code

* fix yaml bugs

* add set dtype

* move proposal_impl to funcs

* fix compile bugs
```
  8f8ed7de
- R
  [CustomDevice] add profiler apis (#45130) · da51baf2
  由 ronnywang 提交于 8月 23, 2022
```
* [CustomDevice] add profiler apis

* migrate CalculateEstOccupancy into cuda_tracer

* update

* add ut
```
  da51baf2
22 8月, 2022 7 次提交
- J
  Add int8 support for matmul+elementwise_add fuse pass (#45077) · 9e5f3a38
  由 joanna.wozna.intel 提交于 8月 22, 2022
```
* Add int8 support for matmul+elementwiae_add fuse

* Corrections after review and ernie test fix
```
  9e5f3a38
- S
  Extend conv_concat_relu to support all activations (#45089) · d03ef054
  由 Sławomir Siwek 提交于 8月 22, 2022
```
* merge conv_concat_relu to conv_act

* fix typo

* extend unit test

* reuse existing gpd

* codestyle

* enforce mkldnn conv
```
  d03ef054
- Z
  
  [Paddle-TRT] support output_padding in conv2d_transpose and conv3d_transpose (#45004) · 25d25b00
  由 zhoutianzi666 提交于 8月 22, 2022
  
  25d25b00
- W
  [Eager] some python c api use final state (#45221) · d2ef888b
  由 wanghuancoder 提交于 8月 22, 2022
```
some python c api use final state
```
  d2ef888b
- Y
  
  remove trt_skip_layernorm_fuse_pass from gpu passes (#45293) · 25d58db6
  由 Yuanle Liu 提交于 8月 22, 2022
  
  25d58db6
- H
  [jit] add jit layer function default constructor (#45169) · e3574f72
  由 Hui Zhang 提交于 8月 22, 2022
```
* fix jit layer function

* fix comment

* fix comment
```
  e3574f72
- R
  
  [CustomDevice] fix custom ccl (#45276) · 307ad60d
  由 ronnywang 提交于 8月 22, 2022
  
  307ad60d
20 8月, 2022 2 次提交
- W
  [Eager] pylayer detach output tensor if it is equal with input (#45065) · bba13e21
  由 wanghuancoder 提交于 8月 20, 2022
```
* pylayer detach output tensor if it is equal with input

* pylayer detach output tensor if it is equal with input
```
  bba13e21
- S
  【autograd】add max_p primitive operator for new autograd (#45178) · 197f4048
  由 Sing_chan 提交于 8月 20, 2022
```
* add max_p without test

* add test of max_p

* make max_p consistent with paddle.maximum
```
  197f4048
19 8月, 2022 9 次提交

W
fix layernormTrt meanVar alloc bug (#45255) · 6fb34e74
由 Wang Bojun 提交于 8月 19, 2022
```
* fix layernormTrt meanVar alloc bug
```
6fb34e74
H

polish REGISTER_OPERATOR parameter of fill_any (#45263) · 1c4134f6
由 HongyuJia 提交于 8月 19, 2022

1c4134f6
R
Fix random op dependency and lr_shedule bugs for standalone executor (#45265) · 6d4ae007
由 Ruibiao Chen 提交于 8月 19, 2022
```
* Fix random op depenency and lr_shedule bugs for standalone executor

* Fix CI errors

* Fix CI errors

* Fix CI errors
```
6d4ae007
W
Trt groupnorm dynamic plugin (#44911) · 1aa6adb1
由 Wang Bojun 提交于 8月 19, 2022
```
* add group_norm dyanmic plugin
```
1aa6adb1
H

[XPU] c_allreduce support int. update bkcl to 1.0.5. test=kunlun (#45248) · 9f1f1b0a
由 houj04 提交于 8月 19, 2022

9f1f1b0a

Make up beam_search_decode operator test cases on xpu and cpu environment (#45264) · 3deab77f

由 mengqingchun02 提交于 8月 19, 2022

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* make up beam_search_decode operator test cases on xpu and cpu environment. test=kunlun

3deab77f

[XPU] add merged_momentum unittest and change momentum (#45241) · e0f1c9f2

由 dongfangshenzhu 提交于 8月 19, 2022

* add merged_momentum *test=kunlun

* add merged_momentum *test=kunlun

* add fp16 to merged_momentum,*test=kunlun

* change dist_model.cc

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

e0f1c9f2

fix some auto code generation bugs (#45232) · 9556c688

由 Charles-hit 提交于 8月 19, 2022

* 修复生成动态图代码时，如果输出没有配置名字，会导致下标越界的问题。

* decide forward_return[0] is not none

* 修改反向yaml前向输出只有一个时，未配置名字，那么输出自动生成为out

* modify code style

9556c688

Support beam search decode op in XPU environment (#44917) · adaffb7b

由 mengqingchun02 提交于 8月 19, 2022

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

adaffb7b

18 8月, 2022 1 次提交

[inference]predictor add GetInputType interface (#45143) · a8ae87f1

由 heliqi 提交于 8月 18, 2022

* predictor add GetInputType interface

* predictor change GetInputType to GetInputTypes

* predictor add tester

* predictor add tester

* predictor change GetInputType to GetInputTypes

* predictor change GetInputType to GetInputTypes

* predictor add tester

a8ae87f1

机器未来 / Paddle 与 Fork 源项目一致

机器未来 / Paddle
与 Fork 源项目一致