*  add FP16 Pass 

* Support the auto completion of while_op

*  acc aligned

* no

* fix bugs

* fix doc according to review

* fix api doc format

* fix api doc according to review

* fix bug and add unit test

* fix record event bug

* fix code stype

* update unitest

[DoubleGrad PR #1] Decoupled code generation logics for Dygraph ForwardFunctions and GradNodes (#40937)

* [Refactor] refactored eager_gen.py PR #2

* [DoubleGrad PR #1] Decoupled code generation logics for Dygraph ForwardFunctions and GradNodes

* Fixed minor issue

* remove in_dtype, out_dtype in redcue grad

* set the dtype and layout in noneedbufferInputs func

* So Parser

* add macro

* add macro

* slotrecord

* add macro

* code format

* fix outpu dtype of elementwise_div

* fix ci

* fix ci

* add fused_seqpool_cvm op;test=develop

* fix assign kernel bug

* fix xpu kernel select error

* add cudn pinned place

* fix copy error

* fix infrt error

* moving OP

* move forward

* move grad and infershape

* code format

* format code

* fix code

* fix code

* fix CMakerLists.txt

* fix comments

* Refine CMakeLists for rocm ci

* [Eager] Support SelectedRows in eager mode

* Remove unnecessary codes

* Adapt new dygraph flag

* update docs dtype(core.VarDesc.VarType)

* fix code style, test=document_fix

fix code style, test=document_fix
Co-authored-by: NChen Long <1300851984@qq.com>

* move assign value op kernel

* remove original op

* fix failed unittests

* remove infershape

* add adaround post-quant method

Enabled eager_mode for complex unit tests, except for test_complex_op.py and test_complex_view_op.py (#40887)

* perfect reshape kernel

* fix bugs of sig

* add unittest for reshape_sig

* fix bugs when run converage

* [Dy2Stat] Fix ForLoop Transformation with single return

* [Dy2Stat] Fix ForLoop Transformation with single return

* Refine test_lac.py for eager mode

* refine code

* Fix test_program_translator for eager

* Fix bug while specifying target grad in high order gradient

* add more unittest

* add more unittest

* add enum from nvinfer namespace to trt lower pattern

* update

* first version, maybe many errors

* refactor op_test

* fix compare list

* fix bg

* fix bugs

* fit for mkldnn and inplace op

* fix compile

* refine ut

* register op version

* fix inplace op

* fix transfer_layout

* support check data type and mutable_data with compiled infos in paddle with cinn

* update cinn_instruction_run_op_test with multi data type

* move slice to pten

* merge develop; test=develop

* fix slice bug;

* update

* update

* fix error

* update

* fix bug

* polish code

* polish code

* polish code

* try to fix windows bug

* add gpu compile flag;

* try to fix

* remov template;

* polish code;

* fix npu bug;

* fix npu bug

* fix npu bug; test=develop

* fix slice bug;

* remove no need dep

* Make StreamSafeCUDAAllocator compatible with NaiveBestFit strategy

* Set FLAGS_use_stream_safe_cuda_allocator to false

* Update

* Remove unnecessary code

* Fix CI errors

* Add UT

* [NPU] fix npu cast ut

* [NPU] fix npu cast ut

* fix inplace bug in final_state eager_gen

* fix python_c_gen

* fix amp with optiontional api bug

* refine optional code for amp