* allow amp and recompute working together

Weight quantization support channel_wise_abs_max method to achieve higher accuracy (#23629) (#23937)

* Weight quantization support channel_wise_abs_max method to achieve higher accuracy

* Update for QAT INT8 MKL-DNN document, added info on VNNI in Windows, benchmark results added and updated

* 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

* Collect output scale for quantized op and fused op
* Post_training_quantizaion sets batch_generator to support lod tensor

* add tdm child op

* Post_training_quantizaion supports min_max methon

* Added an option to use external FP32 model in QAT comparison test

* fix basic gru lstm parameter attr bug; test=develop

* fix bias attr bug; test=develop

* add basic lstm gru name unitest; test=develop

* add partial_sum_op, test=develop

* modify the Paddle Error Message, test=develop

* modify the Paddle Error Message, test=develop

* modify the bug for python3, test=develop

* modify the ut for ci, test=develop

* mv to contrib, test=develop

* use check_variable_and_dtype, test=develop

* fix ci, test=develop

* fix conflict, test=dvelop

* add partial concat, test=develop

* fix the conflict, test=develop

* fix the error, test=develop

* rm SSE4, test=develop

* add partial_concat, test=develop

* fix the grids and blocks, test=develop

* fix the Paddle_Enforce, test=develop

* fix the doc of op, test=develop

* fix the doc, test=develop

* fix the doc of the op, test=develop

* replace -1 with None, test=develop

* Doc update with Ernie QAT INT8 benchmarking

test=develop

* fixes after review

test=develop

* remove ernie part, test=develop test=document_fix

* Fix model name for qatv2

test=develop test=document_fix

* Add Ernie data

test=develop test=document_fix

* update ERNIE benchmark with baidu QA results, test=develop test=document_fix
Co-authored-by: Nbingyanghuang <33643817+bingyanghuang@users.noreply.github.com>
Co-authored-by: NMichał Gallus <sand3r@interia.eu>

* a test for Ernie QAT INT8 accuracy check

test=develop

* Remove NLP comparison test to split PRs

test=develop

* Fix typo and tabs, delete commented lines

test=develop

* re-combine the 2 PRs, test=develop
Co-authored-by: NMichał Gallus <sand3r@interia.eu>
Co-authored-by: Nbingyanghuang <33643817+bingyanghuang@users.noreply.github.com>

* post_training_quantization support set bits, test=develop

* up, test=develop

* support weight quantization in post_training_quanzitaion, test=develop
* add test for weight quantization, test=develop

* add mul and matmul quantization, test=develop
* add test for matmul, test=develop

* use snake_cased Layer name

* add named_parameters and named_sublayers api

* add include_sublayers param, add unittest
test=develop

* fix named unittests
test=develop

* fix unittest
test=develop

* add api docs
test=develop

* arg fix
test=develop

* reserve rnn_impl name_scope for static graph
test=develop

* fix load static param
test=develop

* fix load static param
test=develop