提交 · c9285a18a0e75d20adde189e77afba86b43d5155 · 机器未来 / Paddle

28 7月, 2020 2 次提交
- Y
  saving inference model when user define activation or weight preprocess function (#25749) · c9285a18
  由 yukavio 提交于 7月 28, 2020
```
* saving inference model for user defined quantization model

* saving inference model for user defined quantization model

* fixed ci coverage
```
  c9285a18
- C
  
  Fix test_quantization_scale_pass by change the model, test=develop (#25710) · 650d7223
  由 cc 提交于 7月 28, 2020
  
  650d7223
27 7月, 2020 1 次提交
- W
  Add an option to choose inference targets in Quant tests (#25582) · 43f3d0cc
  由 Wojciech Uss 提交于 7月 27, 2020
```
test=develop
```
  43f3d0cc
15 7月, 2020 1 次提交
- L
  Add bilateral_slice op (#25401) · 7129f544
  由 LielinJiang 提交于 7月 15, 2020
```
* add bilateral slice op
```
  7129f544
13 7月, 2020 2 次提交
- Y
  
  disable unittest test_user_defined_quantization,test=develop (#25451) · 3e45d44d
  由 YUNSHEN XIE 提交于 7月 13, 2020
  
  3e45d44d
- Z
  
  add more unit tests for imperative qat. test=develop (#25486) · ee44bcdd
  由 Zhen Wang 提交于 7月 13, 2020
  
  ee44bcdd
11 7月, 2020 1 次提交

Quantization-aware training for dygraph (#24634) · 548cdbc5

由 Zhen Wang 提交于 7月 11, 2020

* Add the imperative quantization aware training.
*  This is the python part of Imperative QAT. test=develop

548cdbc5

07 7月, 2020 2 次提交
- C
  
  Use the specificed scope in post quant, test=develop (#25384) · 5c8e7995
  由 cc 提交于 7月 07, 2020
  
  5c8e7995
- C
  
  Fix post quant save bug, test=develop (#25370) · 22720a15
  由 cc 提交于 7月 07, 2020
  
  22720a15
06 7月, 2020 2 次提交
- W
  
  Quant2 updates and fixes (#25313) · d0a921ba
  由 Wojciech Uss 提交于 7月 06, 2020
  
  d0a921ba
- C
  [Quantization] Save output threshold by argname_index (#25272) · d8f4714b
  由 cc 提交于 7月 06, 2020
```
* Save output threshold by argname_index, test=develop
```
  d8f4714b
30 6月, 2020 1 次提交
- W
  rename qat into quant (#24948) · 23a4f54b
  由 Wojciech Uss 提交于 6月 30, 2020
```
test=develop
```
  23a4f54b
29 6月, 2020 1 次提交
- W
  rename qat into quant in filenames only (#25194) · 56fa3880
  由 Wojciech Uss 提交于 6月 29, 2020
```
test=develop
```
  56fa3880
19 6月, 2020 1 次提交
- I
  
  disable unitest for gcc8(#25134) · f2825992
  由 iducn 提交于 6月 19, 2020
  
  f2825992
17 6月, 2020 1 次提交
- C
  
  Support conv2d_traspose quantize, test=develop (#25084) · 8fc31d50
  由 cc 提交于 6月 17, 2020
  
  8fc31d50
04 6月, 2020 1 次提交

support user defined quantization func and preprocess (#24720) · b174b997

由 Liufang Sang 提交于 6月 04, 2020

* add user defined func test=develop

* update test=develop

* update test=develop

* fix name conflicts test=develop

* add unittest test=develop

* change 2018 to 2020 test=develop

* add comment test=develop

* add comment for function test=develop

* fix details test=develop

* fix details test=develop

b174b997

02 6月, 2020 2 次提交
- C
  Post training quantization supports optimize model by fusing (#24822) · 75eec3d1
  由 cc 提交于 6月 02, 2020
```
* Post_training_quantization supports optimize model by fusing, test=develop
```
  75eec3d1
- W
  
  add option to exclude ops by id from quantization (#24689) · 78d4f0cc
  由 Wojciech Uss 提交于 6月 02, 2020
  
  78d4f0cc
27 5月, 2020 1 次提交
- C
  
  Update sigmoid output from Y to out, test=develop (#24765) · dbcd7c69
  由 cc 提交于 5月 27, 2020
  
  dbcd7c69
26 5月, 2020 3 次提交
- C
  
  Collecting concat output threshold, test=develop (#24742) · 88e9d74a
  由 cc 提交于 5月 26, 2020
  
  88e9d74a
- S
  
  fix conflict, test=develop (#24238) · 95089204
  由 ShenLiang 提交于 5月 26, 2020
  
  95089204
- C
  
  Add output threshold for ops that have several output activations, test=develop (#24726) · 6c89ca21
  由 cc 提交于 5月 26, 2020
  
  6c89ca21
14 5月, 2020 1 次提交
- L
  Update DNNL QAT document 2.0-alpha (#24494) · 8ef3c02e
  由 lidanqing 提交于 5月 14, 2020
```
Update DNNL QAT document 2.0-alpha
```
  8ef3c02e
13 5月, 2020 1 次提交

[Fix bug] Init scale node in OutScaleForTrainingPass and enable... · 4d351122

由 cc 提交于 5月 13, 2020

[Fix bug] Init scale node in OutScaleForTrainingPass and enable test_quantization_scale_pass UT (#24393)

* Init scale node in OutScaleForTrainingPass, test=develop
* Enable test_quantization_scale, test=develop

4d351122

12 5月, 2020 1 次提交
- J
  
  Model converter to dot file (#23169) · 53125c2f
  由 joanna.wozna.intel 提交于 5月 12, 2020
  
  53125c2f
08 5月, 2020 1 次提交
- W
  Enabled quantize all and skip missing in QAT (#24281) · db052009
  由 Wojciech Uss 提交于 5月 08, 2020
```
* Enabled quantize all and skip missing in QAT
```
  db052009
29 4月, 2020 1 次提交
- L
  
  add try finally, test=develop (#24243) · 381492fc
  由 Leo Chen 提交于 4月 29, 2020
  
  381492fc
28 4月, 2020 2 次提交

Update QAT INT8 2.0 doc (#24127) · 61ec30f0

由 lidanqing 提交于 4月 28, 2020

* update local data preprocess doc

* update for 2.0 QAT

test=develop

test=document_fix

* update benchmark data

test=develop

test=document_fix
Co-authored-by: NWojciech Uss <wojciech.uss@intel.com>

61ec30f0

S

added reshape transpose matmul fuse pass (#23754) · e1a7a880
由 Sylwester Fraczek 提交于 4月 28, 2020

e1a7a880

26 4月, 2020 1 次提交

Add batch_fc op in contrib (#24017) · 0fb9b208

由 ShenLiang 提交于 4月 26, 2020

* add batch fc op, test=develop

* add batch_fc_op, test=develop

* fix untest, test=develop

* rm check_dygraph, test=develop

* fix comment, test=develop

* fix comment, test=develop

0fb9b208

24 4月, 2020 1 次提交
- A
  
  added fusing matmul-transpose-reshape pass (#23866) · d31a174f
  由 arlesniak 提交于 4月 24, 2020
  
  d31a174f
23 4月, 2020 1 次提交

QAT: support for new models (#23928) · 3d744162

由 Wojciech Uss 提交于 4月 23, 2020

* QAT: support range-based quantization and scales from attribute

* added support for channelwise

3d744162

20 4月, 2020 2 次提交

Z

remove high level api (#23854) · 6bd200db
由 zhangchunle 提交于 4月 20, 2020

6bd200db

Add rank_attention_op attributes for GPU memory in contrib (#23915) · 30bd7e1c

由 ShenLiang 提交于 4月 20, 2020

* optimize rank_attention, test=develop

* use the paddle memory pool, test=develop

* set max size, test=develop

* limit the max size, test=develop

* fix the head of cu, test=develop

* add AsDispensable, test=develop

30bd7e1c

17 4月, 2020 1 次提交
- C
  Weight quantization support channel_wise_abs_max method to achieve higher accuracy (#23629) · 40aa14ec
  由 cc 提交于 4月 17, 2020
```
* Weight quantization support channel_wise_abs_max method to achieve higher accuracy
```
  40aa14ec
15 4月, 2020 1 次提交
- M
  fix AMP and recompute (#23551) · f0e743f1
  由 mapingshuo 提交于 4月 15, 2020
```
* allow amp and recompute working together
```
  f0e743f1
13 4月, 2020 1 次提交
- J
  
  Add scale-matmul fuse pass (#23734) · 12ba05ce
  由 joanna.wozna.intel 提交于 4月 13, 2020
  
  12ba05ce
11 4月, 2020 1 次提交
- W
  [Doc update] Update for QAT INT8 MKL-DNN document (#23361) · 2383a9f7
  由 Wojciech Uss 提交于 4月 11, 2020
```
* Update for QAT INT8 MKL-DNN document, added info on VNNI in Windows, benchmark results added and updated
```
  2383a9f7
10 4月, 2020 2 次提交

Add Tdm sampler op in Contrib (#23290) · 8c0bdde9

由 Chengmo 提交于 4月 10, 2020

* add tdm sampler op

* fix compile bug

* fix compile bug

* fix compile bug

* fix compile bug

* test=develop, add tdm sampler unittest

* fix tdm sampler unittest

* fix input var name bug

* update tdm sampler unittest

* fix unittest

* fix unittest

* update tdm sampler unittest

* add tdm exe run unittest

* fix infershape

* test=develop, add doc

* test=develop, fix gcc8 compile bug & unittest bug

* test=develop, fix unittest

* test=develop, fix T one & zero

* test=develop, add unittest check

* test=develop, add doc smaple code & fix dype set

* test=develop, fix dtype

* test=develop, fix compile bug

* test=develop, fix unittest

* test=develop, fix unittest

* test=develop, check py3 unittest

* test=develop,fix unittest

* test=develop, fix py3&py2 unittest diff

* test=develop, fix sample code

* test=develop, fix sample code

* test=develop, fix sample code

* test=develop, fix error message

8c0bdde9

W

Enable matmul and cleanup in QAT2 (#23657) · 1753860d
由 Wojciech Uss 提交于 4月 10, 2020

1753860d

机器未来 / Paddle 与 Fork 源项目一致

机器未来 / Paddle
与 Fork 源项目一致