1. test_parallel_executor_seresnext_base_gpu failed on 2 P100 GPUs with `470.82` driver.
```
======================================================================
FAIL: test_seresnext_with_learning_rate_decay (test_parallel_executor_seresnext_base_gpu.TestResnetGPU)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/opt/paddle/paddle/build/python/paddle/fluid/tests/unittests/test_parallel_executor_seresnext_base_gpu.py", line 32, in test_seresnext_with_learning_rate_decay
    self._compare_result_with_origin_model(
  File "/opt/paddle/paddle/build/python/paddle/fluid/tests/unittests/seresnext_test_base.py", line 56, in _compare_result_with_origin_model
    self.assertAlmostEquals(
AssertionError: 6.8825445 != 6.882531 within 1e-05 delta (1.335144e-05 difference)
----------------------------------------------------------------------
```
2. To be more accuracte on evaluating loss convergence, we proposed to apply IOU as metric, instead of comparing first and last loss values.
3. As offline discussion, we also evaluated convergence on P100 and A100 in 1000 interations to make sure this UT have the same convergence property on both devices. The curves are showed below.
![A100-Single, P100-Single and Diff (1)](https://user-images.githubusercontent.com/13541238/175461920-25df6101-6dd8-4387-862c-d1c8e9299c57.png)

* [Sparse]add SparseTensor mv kernel(csr*dense_vec->dence_vec, coo*dense_vec->dense_vec)

* fix CI

* inplace op without backward supports input of multiple inplace vars

* support inplace without backward in eager mode and fix python_c in eager_fluid

* fix output in yaml

* fix map to const&

* solve conflict v2

* enable Jacobian,Hessian supporting new autograd

* fix prim mode failed in PR-CI-Windows

* add forward_gradients api

* add forward_gradients api

* skip test_autograd_functional_prim in windows ci

* fix test_autograd_funciton_prim timeouot

* remove the block parameter in prim2orig method

* remove duplicate to_tensors code snippet # test=allcases

* [Dy2Stat]Refactor convert_shape transformer logic

* clean usless unittest

* rename eagerpylayer

* move apis from legacy_api.yaml to api.yaml

* remove comment

* revert argsort

* fix merge conflict

* [CustomDevice]add custom place supports

* sync format

* Update kaiming.py

* Update initializer.py

* fix doc bug;test=document_fix

* fix doc;test=document_fix

* Update initializer.py

* Update kaiming.py

* for ci;test=document_fix
Co-authored-by: NLigoml <39876205+Ligoml@users.noreply.github.com>

use jinja 2.11.3 in python lower than 3.6.2

* tmp fix

* init

* compile ok

* compile ok

* add vlogs

* add test

* fix termination error

* add testfile

* add

* fix window compile

* fix window compile

* fix windows compile

* fix windows compile

* fix windows compile

* fix windows compile

* fix windows compile

* fix windows compile

* fix kunlun compile

* fix compilation

* fix compilation

* fix compilation

* tmp fix

* add windows

* add windows

* add more logs

* change timeout to protected

* SB

* add

* add

* fix timeout

* add

* fix test

* fix test

* fix test

* fix ut

* fix ut

* fix ut

* add closure analysis for control flow and add some unittest

* finetune the design of FunctionScopeVisitor

* fix

* fix python check

* fix code by code review

* add slice plugin int32 support

* modify xpu unittest to support fp64, *test=kunlun

* modify xpu unittest to support fp64 for KL2, *test=kunlun

* modify xpu unittest to support fp64, *test=kunlun

* modify xpu unittest to support fp64, *test=kunlun

* fix incompatible error

* rmeove default constructor

* add macro

* fix cpu make error

* add DefaultGPUPlace api

* [Auto Parallel] Use a fast completion for data parallelism

* remove unuse cuSparse function

* [Auto Parallel] Fix some bugs of the fast dp completion

* [Auto Parallel] Add the cmake statements

* [Auto Parallel] Make the unittest adapt to the new interface

* [Auto Parallel] Modify the timeout of the unittest

* [Auto Parallel] Remove unnecessary comments
Co-authored-by: Nzhouwei25 <zhouwei25@baidu.com>

* move trace into api.yaml

* add trace unittest

* fix trace test

* fix generate op

* rewrite all code examples, test=document_fix

* refine arguments, test=document_fix

* fix desc format error, test=document_fix

* capitalize the first letter, test=document_fix

* refine api desc, test=document_fix

* fix wrong COPY-FROM label in Model docs, test=document_fix

* refine returns, test=document_fix

* refine returns, test=document_fix

* add a blank line in code block, test=document_fix

* add outputs for code examples, test=document_fix

* missing code example update, test=document_fix

* missing returns, test=document_fix

* fix the case of proper nouns, test=document_fix

* fix SqueezeNet code example (add a blank line), test=document_fix

* fix ShuffleNetV2 args style, test=document_fix

* missing paper links, test=document_fix

* missing `optional`, test=document_fix

* refine parameter `pretrained`, test=document_fix

* add parameter `kwargs`, test=document_fix

* refine parameter `num_classes`, test=document_fix

* capitalize the first letter, test=document_fix

* missing output of code example, test=document_fix

* fix wrong api label, test=document_fix

* refine returns, test=document_fix

* add missing `.` after paper link, test=document_fix

* remove extra `:`, test=document_fix

* :math:\`0.0\`, test=document_fix

* [Dy2Stat]Support nonlocal mechanism in IF ast transformer

* support prune return vars in cond

* fix unittest

* fix unittest

* fix static check

* add cast trt converter