* add recompute based checkpoints methods for large batch training
test=develop

* add append_backward_with_forward_recomputation
test=develop

* refine optimizer
test=develop

* update backward and optimizer
test=develop

* make Variable usable
test=develop

* add recompute code

* refine optimizer
test=develop

* refine addup _append_backward_ops_with_checkpoints_
1) for recompute part, just cache the grad_op_desc without appending to block
2) before appending grad_op_desc to backward part, addup_repetitive_vars, remove unused branch
test=develop

* make method private

* add recompute strategy into DistributedStrategy
test=develop

* checkpoint version3
test=develop

* remove some print information
test=develop

* remove unused sumop
test=develop

* try to fix recompute with graph building modules

* add input names to vars should be held

* add memory debug tool

* backup backward

* Fix bugs

* add backward desc for op not in any segments

* add exception info for sub_block

test=develop

* modify code style

test=develop

* modify code style

test=develop

* remove print functions

test=develop

* add API spec

test=develop
test=document_preview

* make Recompute a child class of Optimizer

test=develop
test=document_preview

* add API spec

test=develop
test=document_preview

* modify API spec

test=develop
test=document_preview

* add document for Recompute

test=develop
test=document_preview

* change API doc of Rcompute

test=develop
test=document_preview

* code cleaning

test=develop
test=document_preview

* modify API spec

* fix bugs when segments hold no element

* add testcase for Recompute Optimizer

test=develop
test=document_preview

* add test for apply_gradient, and code cleaning

test=develop
test=document_preview

* add test case for load function

* enable CI

test=develop
test=document

* add test case

test=develop
test=document_preview

* add sample code for 4 function of recompute optimizer

test=develop
test=document_preview

* Add RecordHistoryLocalExecScopes
test=develop

* remove the useless warning for user to avoid confuse test=develop

* add mse_loss op

* add op compatible infomation; test=develop

* add enum type

* add enum type; test=develop

* fix softmax ce time limit check failed. test=develop

* refine softmax calc. test=develop

* move tree_conv to fluid.contrib.layers

test=develop

* update API.spec for tree_conv

test=develop

* update tree_conv api to increase unit coverage

test=develop

* unify DataLoader APIs, test=develop

* integrate iterable CPU Dataset, test=develop
add GPU dataset supporting, test=develop

* add unittests for dataset, test=develop

* add more docs to dataloader apis, test=develop, test=document_preview

* refine doc, test=develop

* refine doc again, test=develop

* increase coverage, test=develop

* add instance norm op

* refine reallocate of workspace size, test=develop

* add lock to cudnn handle calls, test=develop

test=develop

* Initial, functional commit

* Clean commit related files
test=develop

* refactor dygraph,test=develop

* fix failed unittest,test=develop

* polish code,test=develop

* check windows ci error,test=develop
try to fix windows ci error by np.allclose,test=develop

* polish vlog and profiler, test=develop

* try to fix preceding ops order,test=develop

* test transformer in windows ci, test=develop

* use python c-api to speed up tracer.trace,test=develop

* test=develop, fix docker with paddle nccl problem

* test=develop, add ut for debug string and gradient_accumulator

* test=develop, add tests for layer/gradient_accumulator/prepared_op

* test=develop, fix complie error for test_prepared_op

* test=develop, add more ut for dygraph

* test=develop, create API.spec for dygraph api change

* test=develop, refoctor name to make it easier to understand

* test=develop, refoctor name to make it easier to understand

* test=develop, fix multi-gpu failed problem , add Tracer tests, change PADDLEENFORCE to PADDLEENFORCE_EQ

* test=develop, fix ut failed on parallel se-resnext

* test=develop, change one more PADDLE_ENFORCE

* support auto prune in dygraph mode

* test=develop, support auto prune

* test=develop, merge develop conflict

* test=develop, fix test_layer and test_tracer ut

* test=develop, fix bug which may cause stop_gradient disabled with a list of backward inputs

* add TRT shape check, test=develop

* model_input_shape == runtime_input_shape, refine message, test=develop

* fix trt bugs when sharing params, test=develop

* add unittest for cascade_rcnn

* refine err msg of out of gpu memory, test=develop

* refine err msg again, test=develop

* refine errog message again, test=develop

* follow reviewer's comments, test=develop

* support 2-level lod of input in sequence_pool test=develop

* fix lod level bug in .cu test=develop

1. group_norm support data_layout=NHWC
2. modified doc of group_norm

* Set states of recurrent op as dependent vars in prune of save inference model

This PR will fix the save/load inference model problem of RNN models.

The reason of the bug is that save_inferenc_model will prune OPs that doesn't contribute to Output. But in recurrent_op, States are not Output, OPs refers States will be pruned. 

This fix adds States of recurrent_op as dependent var so that OPs referring States won't be pruned. 

- LRN mkl-dnn kernel refactor

test=develop

- compilation fix

- Another compilation fix

- Compilation fix

- another compilation fix

- compilation fix

- Crash fix

- optional LRN mkldnn workspace

- Added mid allocation

- Workaround for tests

- Removed gradient from is_test ut

- Removed mid for inference

- Reverted LRN mid removal for is_test

- PADDLE_ENFORCE adjusted

- Rebase to templatization commit

- Compilation fix

- compilation fix

test=develop

- lint

test=develop

- Fix to crash

- Rebase to recent codebase

 - lin

- lint

- compilation fix

modified interpolate_op to support tensor attribute

1. the parameter out_shape of image_resize、resize_nearest/bilinear/trilinear can be a list or a 1-D tensor variable. If a list, each element can be an integer or a tensor variable with shape: [1].

2. the parameter scale of above Ops can be a 1-D tensor variable.
modified document of image_resize, resize_nearest, resize_bilinear, resize_trilinear and add some code example.

add crop_tensor op. The main difference with crop is :

1. If the argument shape is a list, each element is an integer or a tensor variable with shape: [1]. This way is suitable for the case that the shape may be changed each iteration.

2. If the argument shape is a variable. Its rank must be 1. In crop op, the rank of shape must be the same as x

offsets can be a list, in which each element is an integer or a tensor variavle with shape: [1].

* fix conflicts
test=develop

* change mask_bias_reorder
test=develop

* add ComputeMask function to make code clear
test=develop

* change according to reviews
test=develop

* change according to reviews
test=develop

* Fix conv2d+dequantize squash for residual fusion

test=develop

* Change condition

test=develop

Add boost as dependency of prune

fix #19862

* Add template functions for Acquire primitive/primitive_desc
test=develop

* Move acquire primitive descriptor to protected section
test=develop

* Remove constraint that last dimension is forced to be 1 in cross_entropy
test=develop

* modify labels last dims test=develop

* add precise roi pooling op test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* test=develop

* detail the description test=develop

* test=develop

* elaborate the doc for return type test=develop

* test=develop

* Add fc_elementwise_layernorm_fuse pass and unittest.

* Add fused_fc_elementwise_layernorm op and its GPU kernel.
test=develop

* Apply fc_elementwise_layernorm_fuse_pass to GPU inference.

* Add the setting of attrs in the definition of binary_op.
test=develop

* Add comment.

* Implement the unittest.
test=develop

* Change the unittest name of layer_norm.
test=develop

Optimize amp for multi-gpu to enable FP16 gradients transfer across gpus

* strided_slice op basic function test=develop

* test=develop rewrite and fix

* fix bug test=develop

* fix for the PADDLE_ENFORCE usage

* add some unit testw

* fix for the aip  test and copright and fix test=develop

* fix API.spec test=develop

* fix API.spec test=develop

* add axis parameter test=develop

* fix for the build error test=develop

* fix python api  test=develop

* fix the build test=develop

* fix build test=develop

* fix API spec test=develop

* test=develop add some comment and single op test

* fix API spece test=develop

* fix test=develop

* fix test=develop

* fix api test=develop

* fix api test=develop

* fix API.spec test=develop

* fix typo test=develop

* fix API.spec test=develop

* fix API typo test=develop

* fix doc and API.spec test=develop