* Saving inference model for user defined quantization

* Saving inference model for user defined quantization

* 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
Co-authored-by: NLiufang Sang <slf12thuss@163.com>

* Update sigmoid output from Y to out, test=develop (#24765)

* Collecting concat output threshold, test=develop (#24742)

* Add output threshold for ops that have several output activations, test=develop (#24726)

* [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

test=release/2.0

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

* added support for channelwise

test=release/2.0

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

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

* Post_training_quantizaion supports min_max methon

* 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

* fix post training quantization bug of memory constrained, support the input be different, test=develop

* add ut for comparing FP32 and QAT INT8

* add save qat transformed model python script
test=develop

* updated

* added missing file

* add "with_label"
test=develop

* performance benchmark as unit test
test=develop

* change names of unnecessary thing

* Change CMakeList.txt for model downloading and UT
test=develop

* change names of functions and params for more readable code
test=develop

* Change PADDLE_ENFORCE messages
test=develop

* fix indent problems
test=develop

* indent problems
test=develop

* support set model_filename and params_filename in post_training_quantization, test=develop

* Support  more ops in post training quantization, and save the output scale in quantized op.
* Update docs in post training quantization and qat 

test=develop

* Fix bug for inserting add_quant_dequant_op to same variable repetitively in add_quant_dequant_pass, test=develop

* add post training quantization, test=develop
* specify the quantizable op type, test=develop

* move pool2d to add_quant_dequant_pass, test=develop

* fix fuse_reduce_op quantization bug test=develop

* close fuse_all_reduce_ops in PaddleSlim, test=develop

* Leave fake quantization around mul

* Replace Fake with Real Quantized Mul

* Gather all scales from fake_quantize_ops

* Enable uint8 in conv_relu tensors

* Disable int8 mul and restore fake mul

* Fix buf for running QAT on VGG16 and 19

* Follow Wangzhen's comment in PR 18970, test=develop

* Review comments, test=develop

* Leave fake quantization around mul

test=develop

* Replace Fake with Real Quantized Mul

test=develop

* Fix bug in quantize placement pass

Nodes in the graph now have checked type instead of node name when they are to be marked for quantization test=develop

* Add support for new QAT models

test=develop
Co-Authored-By: NMichał Gallus <michal.gallus@intel.com>
Co-Authored-By: NWojciech Uss <wojciech.uss@intel.com>

* fixed fps results

test=develop

* fix top5 accuracy drop problem

* updated for new QAT models

* skip quantizing average pooling - dirty but working

* add missing pass

* added missing conv+brelu fuse pass

* removed a call to non-existent pass

test=develop

* renamed pass

test=develop

* Adjust finding pooling scale to newest QAT models

* Remove unnecessary code from quantization_mkldnn_pass

* Copy Pooling input scale to output scale in QAT

* Refactor & remove unused code in QAT

* Incorporate fp32 FC into QAT

test=develop

* Enable graph drawing with debug flag

test=develop

* Add tests for QATv2

* Fix paths for QATv2 models

test=develop

* Add option to save transformed int8 qat model

test=develop

* Remove redundant lines from qat mkldnn pass

test=develop

* Delegate disablement of avg pooling to qat

test=develop

* fix CI bug, test=develop

* Follow Wangzhen's Review, test=develop

* Update API.spec

test=develop

* Name False in (is_unsigned, TensorScale) tuple

test=develop

1. Support customize eval function instead of eval program.
2. Fix loading checkpoint in quantization strategy.
3. Support saving eval model when saving a checkpoint.
4. Fix decoder of loading context in PaddleSlim.
5. Fix restoring from the checkpoint of uniform prune strategy.
6. Support saving eval model and infer model during training.
7. Add ‘unitest’ for saving eval model, saving infer model and uniform pruning restoring from the checkpoint.
8. Fix pruning of depthwise_conv_grad op by updating the groups.

* add fake_quant_dequant_op for average pool2d
* add test

test=develop

* python inference api supports numpy and add unit test, fix unit test fail in test_slim_int8_googlenet and test_slim_int8_mobilenet

* add pool2d quantization support, only for max-pooling.

* add the partial quantization support.