提交 · dfed4a63f2334d6e4034b583e17085fa9b0fe35d · 马赛克医院 / Paddle

09 12月, 2021 17 次提交
- H
  support offload in sharding stage2 (#37904) · dfed4a63
  由 Haohongxiang 提交于 12月 09, 2021
```
* merge latest develop branch

* fix bugs

* update

* fix bugs for unittest

* modify for less use of gpu mem

* fix bugs of using _reset_grad_inplace_version

* update

* update

* modify for CI-Coverage

* retrick all CIs
```
  dfed4a63
- H
  add transpose_flatten_concat_fuse_pass test case (#37675) · a9f81534
  由 heliqi 提交于 12月 09, 2021
```
* add transpose_flatten_concat pass

* modify skip func to ignore_pass_case func

* delete input_shape limit

* modify get node order
```
  a9f81534
- Y
  
  [fleet_executor] set learning rate for each run and set op_role for fetch var (#37945) · 59bd1e6f
  由 Yuang Liu 提交于 12月 09, 2021
  
  59bd1e6f
- C
  
  fix kernel selected error caused by alias name (#37984) · f68b175f
  由 Chen Weihang 提交于 12月 09, 2021
  
  f68b175f
- A
  
  fix LoDTensorArray crash in Debug mode build (#37954) · 1f9a5d8f
  由 Aganlengzi 提交于 12月 09, 2021
  
  1f9a5d8f
- S
  Refine CUDA atomicAdd for FP16 by CUDA primitive methods (#37895) · 033ebe7e
  由 sneaxiy 提交于 12月 09, 2021
```
* fix cuda atomicAdd for FP16

* try to fix ci
```
  033ebe7e
- P
  
  fix Uniform sample method (#37823) · 491d4f01
  由 pangyoki 提交于 12月 09, 2021
  
  491d4f01
- Z
  Fixed eager compilation issues by temporarily turn off AutoCodeGen fo… (#37992) · 34a06cf5
  由 Zhanlue Yang 提交于 12月 09, 2021
```
* Fixed eager compilation issues by temporarily turn off AutoCodeGen for specific ops

* Removed op_types
```
  34a06cf5
- J
  
  add ipu device p2 (#37840) · cb636a48
  由 jianghaicheng 提交于 12月 09, 2021
  
  cb636a48
- R
  
  optimize flip op, removing duplicated computation when dim size is one (#37825) · 890638cf
  由 Roc 提交于 12月 09, 2021
  
  890638cf
- F
  
  format softmax forward (#37927) · 18aca3f5
  由 Feng Xing 提交于 12月 09, 2021
  
  18aca3f5
- B
  
  Add varbase init name (#37947) · fdf62e1e
  由 Baibaifan 提交于 12月 09, 2021
  
  fdf62e1e
- W
  default accessor and multi table config (#37714) · a9e0d28c
  由 wangguanqun 提交于 12月 09, 2021
```
* default accessor and multi table config

* add unittest

* add unittest

* delete print
```
  a9e0d28c
- W
  [Paddle-Inference] fix_ele_convert: IElementWiseLayer can broadcast (#37908) · f695dc97
  由 Wangzheee 提交于 12月 09, 2021
```
* fix_ele_convert: IElementWiseLayer can broadcast

* fix_ele_convert
```
  f695dc97
- C
  
  adjust main dir (#37916) · 1911b6f0
  由 Chen Weihang 提交于 12月 08, 2021
  
  1911b6f0
- C
  
  fix make error by alias name change (#37971) · e3f68f42
  由 Chen Weihang 提交于 12月 08, 2021
  
  e3f68f42
- Z
  Adjusted Eager AutoCodeGen to Support Operators with Multiple OpBases & Enable... · 9aed9ea0
  由 Zhanlue Yang 提交于 12月 09, 2021
```
Adjusted Eager AutoCodeGen to Support Operators with Multiple OpBases & Enable Passing Output Tensor as Input Argument (#37943)

* Rearranged Eager AutoCodeGen directory structure

* Removed USE_OP in Eager AutoCodeGen

* Enabled generation for Operators without Grad/Inputs/Outputs

* Resolved operators without input

* Fixed merge conflicts

* Enabled Eager AutoCodeGen for 10+ more operators

* Refactored Eager AutoCodeGen with more organized helper objects

* Enabled Eager AutoCodeGen for operators with multiple OpBases

* Adjusted Eager AutoCodeGen to Enable Passing Output Tensor as Input Argument
```
  9aed9ea0
08 12月, 2021 21 次提交

[Eager] coreops to 495 (#37926) · aff7397b

由 wanghuancoder 提交于 12月 08, 2021

* refine a test case, test=develop

* publish python c api for eager, test=develop

* revert modify about test_allclose_layer.py, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* delete numpy includes, use pybind11 numpy.h, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* suport eager error msg, and add grad test case, test=develop

* refine, test=develop

* refine, test=develop

* generate eager core ops, only 4 ops, test=develop

* Rearranged Eager AutoCodeGen directory structure

* Removed USE_OP in Eager AutoCodeGen

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* open 500 list

* refine, test=develop

* refine, test=develop

* refine, test=develop

* fix auto code gen, test=develop

* Enabled generation for Operators without Grad/Inputs/Outputs

* refine, test=develop

* refine, test=develop

* refine, test=develop

* add to pyobject, test=develop

* Resolved operators without input

* merge pr 37837

* refine

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine,test=develop
Co-authored-by: Njim19930609 <jim19930609@gmail.com>

aff7397b

Z

[new api]fix rot90 eng doc (#37935) · b72e6021
由 zmxdream 提交于 12月 08, 2021

b72e6021
C
Revert "pylayer support tuple/list type args (#37727)" (#37956) · d848ff04
由 chentianyu03 提交于 12月 08, 2021
```
This reverts commit a73064f2.
```
d848ff04

add a subdirectory named cinn in operators and move releated files into it (#37938) · 9cb637ed

由 CtfGo 提交于 12月 08, 2021

1. add a subdirectory named `cinn` in `paddle/fluid/operators` directory and move releated files into it
2. seperate CinnLaunchContext class from `cinn_launch_op.h` and put it in a  new independent file named `cinn_launch_context.h`, so that it can be included by others clearly.

9cb637ed

Fix host event recorder (#37944) · 20471de7

由 liutiexing 提交于 12月 08, 2021

* add align for WorkQueue

* add spinlock

* merge develop

* merge

* Add EventsWaiter

* Revert "Add EventsWaiter"

This reverts commit e206173aa9be7401b83a53581627bfaf557c8fb2.

* Fix RecordEvent
Co-authored-by: Nliutiexing <liutiexing@google.com>

20471de7

T

Delete git useless files (#37952) · 5cecfd23
由 tianshuo78520a 提交于 12月 08, 2021

5cecfd23
Y
[PTen]Add alias kernel name (#37881) · ff6507db
由 YuanRisheng 提交于 12月 08, 2021
```
* add alias kernel name

* modify code as suggestions
```
ff6507db

Add paddle.lerp API to do a linear interpolation (#37253) · 1716324c

由 wuhuanzhou 提交于 12月 08, 2021

* save temp

* add unittest, test=develop

* fix ci error, test=develop

* fix grad accuracy error, test=develop

* fix unused error, test=develop

* fix compilation error on Windows, test=develop

* add unittest, test=develop

* modify by review comment and add lerp_

* fix inplace api, test=develop

* fix inplace api, test=develop

* fix coverage error, test=develop

1716324c

C
add update func of auto search (#37867) · 46212b80
由 caozhou 提交于 12月 08, 2021
```
* add update func of auto search

* update unitest
```
46212b80
W

[fleet_executor] Add interceptor gc (#37889) · 6b48dfe9
由 WangXi 提交于 12月 08, 2021

6b48dfe9
Y

[fleet_executor] bug fix for fleet_executor, test=allcase (#37934) · 55b87742
由 Yuang Liu 提交于 12月 08, 2021

55b87742
C
implementation of broadcast sub backward by reduce (#37754) · 567e6bbc
由 crystal 提交于 12月 08, 2021
```
* add boardcast_sub

* add boardcast_sub
```
567e6bbc

Fix CUDAGraphAllocator bug for StreamSafeCUDAAllocator (#37821) · b4a67491

由 From00 提交于 12月 08, 2021

* Fix CUDAGraph bug for StreamSafeCUDAAllocator

* Add CUDAGrapthAllocator check in multi-stream interface

* Set FLAGS_use_stream_safe_cuda_allocator defaulted to false

* Fix environment error for cmake

* Fix cmake error

* Add UT of GetAllocatorInterfaceTest

* Add UT of CUDAGraphExceptionTest

* Enhance CUDAGraphExceptionTest

b4a67491

C

add check whether tensor is inplace and leaf when calcute gradient (#37931) · 2c02a580
由 chentianyu03 提交于 12月 08, 2021

2c02a580
F
fix: when ceil_model==true && Padding_algo!=SAME, (x-size)/stride != int, this... · d1ab323f
由 feng_shuai 提交于 12月 08, 2021
```
fix: when ceil_model==true && Padding_algo!=SAME, (x-size)/stride != int, this convert is wrong (#37929)
```
d1ab323f

[Eager] generate eager core ops, only 4 ops (#37813) · 52f63cd2

由 wanghuancoder 提交于 12月 08, 2021

* refine a test case, test=develop

* publish python c api for eager, test=develop

* revert modify about test_allclose_layer.py, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* delete numpy includes, use pybind11 numpy.h, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* suport eager error msg, and add grad test case, test=develop

* refine, test=develop

* refine, test=develop

* generate eager core ops, only 4 ops, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

52f63cd2

Enabled Eager AutoCodeGen for 40+ more operators (#37910) · cf873c39

由 Zhanlue Yang 提交于 12月 08, 2021

* Rearranged Eager AutoCodeGen directory structure

* Removed USE_OP in Eager AutoCodeGen

* Enabled generation for Operators without Grad/Inputs/Outputs

* Resolved operators without input

* Fixed merge conflicts

* Enabled Eager AutoCodeGen for 10+ more operators

cf873c39

Y

fix softmax max dim (#37901) · b5dd12fb
由 Yanxing Shi 提交于 12月 08, 2021

b5dd12fb
S

add pyyaml needed by python\paddle\utils\code_gen\api_gen.py (#37897) · a8f009e4
由 Sing_chan 提交于 12月 08, 2021

a8f009e4
S
Fix CUDA Graph H2D bug by restore host memory (#37774) · a1ad3a63
由 sneaxiy 提交于 12月 08, 2021
```
* fix CUDA Graph H2D bug again

* fix no return bug
```
a1ad3a63
Y

bug fix for adamw (#37905) · 9a2d327c
由 Yuang Liu 提交于 12月 08, 2021

9a2d327c

07 12月, 2021 2 次提交

L

[Fleet Executor] Add feed, fetch and check correctness (#37824) · b8793f70
由 LiYuRio 提交于 12月 07, 2021

b8793f70

introduce INF-RT (#37669) · 70dea138

由 Yan Chunwei 提交于 12月 07, 2021

* add infrt code

refined with Paddle's code style.

* rename CinnRtConfig to InfRtConfig

* rename CinnRt to InfRt of some code

* rename CINNRT to INFRT

* remove unnecessary code

* replace CINN to INFRT in the source code

* replace all "cinn" in code to "infrt"

* remove some const_cast

70dea138

马赛克医院 / Paddle 与 Fork 源项目一致

马赛克医院 / Paddle
与 Fork 源项目一致