* Support pure fp16 training for AMP API. (#29544)

* add cast ops before and after unsupported fp16 ops.

* Keep partial net in FP32 pattern.

* Support check_finite_and_unscale and update_loss_scaling for FP16 calculation mode.

* Add fp16 support for adam op.

* add multi precision attr for adam.

* Fix the bug of test_multi_precision_fp16_train UT.

* Code format for CI.

* Fix the redefine error about MPTypeTrait on windows.

* fix bugs of the _create_accumulators func in Momentum.

* fix bug when inserting post cast op.

* Add the update_loss_scaling op in allow_set of UnusedVarCheck.

* Update for ci coverage.

* Add some doc for OptimizerWithMixedPrecision.

* Fix the code style.

* Imporve the doc of `amp_init`.

* Change for fp16 testing if users have the infer program defined in separate way.

* Remove tensor copy in the update_loss_scaling op. (#29426)

* remove tensor copy in the update_loss_scaling op

* not use thrust.

* fix some cuda memory access error.

* Quantization supports 2.0 APIs

* Fix the error of save_quantized_model

* Optimization grad merge performance (#29784)

* [fleet] combine amp and gradient merge, test=develop (#30086)

* fix assign_op_xpu concat_op_xpu warining (#30120)
Co-authored-by: Nliuyuhui <liuyuhui@baidu.com>

* Optimizer trans momentum (#29597)

* merge amp related function in Momentum from paddle.fluid.contrib.optimizer into paddle.optimizer.

* Add unittest for 2.0  Momentum API.

* fix some bugs in weight_decay.

* add alias for fluid.contrib.mixed_precision (#29562)

* add alias for fluid.contrib.mixed_precision

* add static.amp into setup.pu.in (#29621)

* add static.amp into setup.pu.in

* add unittest for api

* fix a bug in multi_precision_fp16 unittest. (#29756)

* Layer norm fp16 (#29169)

* add fp16 for layer_norm op

* revert layernorm api

* fix forward

* fix forward

* fix backward for layernorm with fp16

* fix unit test for layernorm with fp16

* fix with_mkldnn compile error for layernorm with fp16

* 1. revert to PADDLE_ENFORCE_NOT_NULL, 2. change static_cast<float> to static_cast<U>

* fix with_mkldnn compile error for layernorm with fp16

* fix with_mkldnn compile error for layernorm with fp16
Co-authored-by: Nzhiqiu <chenqiuliang@baidu.com>

* fix layer_norm accuracy (#29434)

* Layernorm opt (#29522)

* layernorm fw opt

* layernorm bw opt

* fix typo, test=develop

* remove const dim3 for windows CI compatibility

* merge develop
Co-authored-by: Nzlsh80826 <zlsh80826@gmail.com>

* Fix compile problem when cuda_arch < 6000 (#29576)

* fix compile problem when cuda_arch < 6000

* refine code

* refine code
Co-authored-by: Nzhiqiu <chenqiuliang@baidu.com>
Co-authored-by: Nzlsh80826 <zlsh80826@gmail.com>

* fix ininite scale values (#29386)

* Support dygraph quant model (#29927)

* Avoid the scale to be infinity in quant2_int8_mkldnn_pass, test=develop
* support quantized model for paddle2.0 dygraph, test=develop
Co-authored-by: NWojciech Uss <wojciech.uss@intel.com>

* [cherry-pick] cherry-pick of PR#29928

* delete paddle.metric.chunk_eval and paddle.metric.mean_iou

* delete paddle.nn.clip and paddle.nn.clip_by_norm

* delete paddle.nn.functional.activation.hard_sigmoid and paddle.nn.functional.activation.hard_swish

* [cherry-pick] cherry-pick of PR#29928

* fix extension import error

* Add pure fp16 training with master weights. (#27712)

* add the weight decay func for the momentum op

* Add the multi_precision function in Momentum Optimizer.

* Make sure that the initial value of master weights are same with the fp16 weights.

* add static loss scaling.

* add the rescale_grad function in the pure fp16 training.

* use the original momentum updating method.

* Polish some codes, such as variable names.

* add docstring for apis.

* update the var creation details of _create_master_weight.

* not modify codes about imperative momentum updating.

* Fix the error of test_dist_sparse_tensor_load_momentum UT.

* add unit test for multi precision fp16 training.

* add more unit tests for CI.

* Use lower threshold values for allclose comparing in test_multi_precision_fp16_train UT.

* Optimiz the unittest test_imperative_out_scale

test=develop

* Impelement 2.0 API version Conv2d and Linear layer quantization in imperative mode.

* use cudnn softmax in static Lenet

* Modified ChannelwiseQAT Unittest for 2.0 API.

* For CI python coverage.

* upgrade comment string to raw string

* fix string in

* fix string with ' '

* revert update on comments

* upgrade only necessary

* fix sample code checker

* fix comments with '''

* refactor momentum op to combine weight_decay (scale op and sum op)

* add debuging code

* change seed & add debug message

* support user-defined quant and preprocess

* add matmtl_v2 to amp list

* support dygraph

* disable ut test_parallel_executor_fetch_isolated_var,test=document_fix

* test for limiting ut exec time as 15S

* fix an error caused by cannot find ut

* fix some error

* can not find test_transformer

* fix error caused by ut not run in windows

* fix error caused by Compiler Options

* fix error caused by setting timeout value as 15 in python/paddle/tests/CMakeLists.txt

* setting timeout value to 120s for old ut

* add the timeout value setting

* fix error caused by ut only run in coverage_ci

* add analyzer_transformer_profile_tester

* fix some error

* fix some error

* fix error with inference option

* fix error with inference option setting as ON_INFER

* add some ut to set timeout

* modified some option

* fix error

* fix some timeout error

* fix error

* fix error

* fix timeout for test_analyzer_bfloat16_resnet50

* fix error

* setting timeout properity for some ut

* first pr for new ut timeout as 15S

* skip reader op in mixed_precision decorator

* add ut

* rename manual_seed to seed

* rename xxx1d-->xxx1D, xxx2d-->xxx2D, xxx3d-->xxx3D

* rename manual_seed --> seed

* do not rename .cc, .cu and .h file

* rename manual_seed --> seed

* rename manual_seed --> seed

* rename manual_seed --> seed

* rename manual_seed --> seed

* disable_static on doc example code

* donot change manual_seed on generator

* add enable_static on sample code

* convert python/paddle/fluid/layers/nn.py to bak

* fix typo

* fix code style

* fix seed to manual_seed when call functions of Generator()

* fix bug

* Implement the function of OueScaleForTraining/OutScaleForInference in dygraph

test=develop

* Add test attribute in channelwise_quant op, test=develop

* replace config by kwargs

* change save path form dir to prefix

* fix failed unittests

* revert unittest name change

* polish en docs

* add more tests for coverage

* fix bilateral bug

* add fused_bn_add_relu op

* default open dygraph mode

* fix CI-Mac

* fix Mac-CI other unittest file

* fix CI-Py3

* fix test_communicator_geo and test_buffer_shared_memory_reuse_pass

* add enable_static to fix CI-Py3

* add enable_static to fix CI-coverage

* delete try except

* Finished ChannelWiseQuantDequantAbsMaxOp and Passed unittests.

* Finished channel-wise quantize strategy in imperative quantization.

* Added Cuda code of ChannelWiseQuantDequantMaxAbsOP
Add Cuda code of ChannelWiseQuantDequantMaxAbsOp

* Add quant_axis for channel_wise quant.

* fixed a bug in unnitests, which will not trigger axis = 1 case and cannot meet the coverage rate requirement.

* Added some assert infomation and fixed some coding style mistakes.

* Remove the cache in post_traning_quantization, test=develop

Update amp_check_finite_and_scale_op and add an updating_loss_scaling op for static graph amp training. (#26240)

* update amp_check_finite_and_scale_op for static_amp.

* use amp_check_finite_and_scale in static graph amp.

* update grads to zero when grads own infinite values(as for amp_checkout_finite_and_scale op).

* add update_loss_scaling op in cpp.

* add update_loss_scaling_op unit test.

* update the doc of the check_finite_and_unscale op

* Update the process of gradients updating skipping if the gradients have infinite values.

* update the way to zero grads.

* update test_update_loss_scaling_op.py

* add log info when find infinite grads.

* add the unit test for UpdateLossScaling Layer.