* use fp32 compute type for cublasGemmStridedBatchedEx with fp16 input/output

* add flags to control compute type

* default to false

* add unit test

* default to true

* add tril_indices cpu kernal

* modify tril_indice cpu op

* modify bug

* modify bug

* add tril_indices python api

* add tril_indices python api

* resolve conflict

* add tril_indices test

* modify details

* add tril_indices.cu

* pythonapi pass

* save tril_indices

* CPU tril_indices pass

* delete vlog

* modify test_tril_indices_op.py

* delete tril_indices_kernel.cc.swp

* delete tril_indice.cu

* modify code style

* add newline in creation.py

* modify creation.py linux newline

* delete annotation

* check code style

* check .py style add final_state??

* modify code style

* add gpu_tril_indices

* modify gpu_compiled_juage

* modify gpu judge

* code style

* add test example

* modify english document

modify english document

modify english document

modify document

modify document

* modify pram name

* modify pram name

* modify pram

* reduce test ex

* change the output format of C++ backward api

* fix merge conflict

* fix sparse api code auto-gen

* fix eager_gen bug

* fix bug of output is null

* fix bug of conv2d_grad_impl

* fix optional grad

* fix bug of eager-gen double_grad

* fix bug

* fix multiply_double_grad bug

* fix bug of higher order derivative

* fix bug of FillZeroForEmptyGradInput

* remove redundant vector in grad_node

* fix bug of test_deformable_conv_v1_op

* fix bug of test_deformable_conv_v1_op

* some refacotr

* remove shared_storage

* fix bug

* fix rnn bug

* refine enforce code

* refine enforce code

* fix compile failed

* fix infrt failed

Add Code Generation for operators,  op makers and argument mapping functions (#41772)

* matmul refactor

* remove UT which only check ENFORCE output

* code format

* improve memory usage

* add intermediate for some api

* fix bug

* fix fluid.layer

* adapt faster tokenizer op

* add eager test

* add unittest

* support construct scalar using non-cpu tensor

* fix bugs when run unittest

* fix compile bugs

* fix bugs when run ci

* fix compile bugs

* fix bugs when move copy

* perfect unit test

* perfect unittest

* update according to comment

* int_array supports constructed by gpu tensor

* add some test

* polish code

* adjust full api

* add unittest

* add unittest
Co-authored-by: NYuanRisheng <yuanrisheng@baidu.com>

* change the output format of C++ backward api

* fix merge conflict

* fix sparse api code auto-gen

* fix eager_gen bug

* fix bug of output is null

* fix bug of conv2d_grad_impl

* fix optional grad

* fix bug of eager-gen double_grad

* fix bug

* fix multiply_double_grad bug

* remove node pruning

* change logic for optimize

* modifty

* Create elementwise_heaviside_op.cc

* add ElementwiseHeavisideFunctor

* Create test_elementwise_heaviside_op.py

* 增加heaviside的python接口

* add heaviside in white list

* 增加heaviside的签名

* 增加heaviside的核函数

* 增加heaviside梯度的核函数

* 增加heaviside梯度的注册

* 调整代码格式

* Update elementwise_sig.cc

* add heaviside in __all__

* Update heaviside docs

* Update math.py

* Update math.py

* Update math.py

* Add AdamW jit kernel

* Second implementation

* Add missing header

* Correct number of jit kernels in the test

* the copy type of data transform for gpu_pinned to gpu change from syna to async

* refactor code

* update xpu depends
*test=kunlun

* minor
*test=kunlun
Co-authored-by: Nroot <root@yq01-sys-hic-p40-0091.yq01.baidu.com>

* fix bug of batch_norm_grad kernel with fp16

* format code

* remove useless lod copy

* fix test failed

* revert meta change

* revert tensor change

* Extend python einsum interface to make einsum_v2 support multi-operands and switch it to default.

* add opt_einsum dependence

* add yaml and support eager model

* fix by code review

* add double yaml

* add inline func

* opt attr eaque perf

* opt attr select code

* fix one hot infermeta

* polish get attr impl

* fix tests failed

* add testcases

* set device id of Place() to get GPUContext needed by LimitGridDim in ElemwiseGradBroadcast

* fix code style

* Fix the race condition in cumsum operator

* Optimize cumsum operator