* add transformer generate square subsequent mask api

* add dtype for input, update doc, use -np.inf

* add dtype for input, update doc, use -np.inf

 optimize the error meesage for detetion_map_op

* numpy -> paddle in conv. test=develop

* Add support for mkldnn ops types selection with FLAGS in dygraph

* use regex to match DNNL verbose

* python3 encoding fix

* set default value to strategy in distributed_optimizer test=develop

* Change docs of WeightNormParamAttr to dygraph model, test=develop, test=document_fix

* remove inplace argument

* fix sample code

* fix sample code

* support not passing block

* fix NumpyArrayInitializer

* add unit test

functionality provided by `normalize`

1. fix a bug that paddle.nn.RNN does not initialize the state for the cell correctly;
2. add unittest for paddle.nn.RNN and paddle.nn.BiRNN

* modified sample code of add_position_encoding to  2.0, test=document_fix

* use core.op in add_position_encoding API.

* add test for add_position_encoding in dygraph mode

* modify doc

* Fix clip_by_norm_op error message. test=develop

* test=develop

* test=develop

* enhance array_to_lod_tensor_op lod_tensor_to_array_op errors information. test=develop

* iCafe 9009: Optimize the error message of OP.

* Optimize the error message of GatherTreeOP.

Remove NLTK requirements to avoid dependency security alert from Github Dependabot

* fix the speed&memory of matmul

* fix the comment

* fix the memory copy

* fix the windows ci

* add unittests and op version register for tensorrt_subgraph_pass

* rename to test_trt_subgraph_pass.py

* fix softmax converter diff when padding dim=1

* test=develop, rm netifaces

* fix error

* fix error

* fix error

* merge develop

* add paddle.fluid._cuda_synchronize, test=develop

* fix bug about core_avx core_noavx, test=develop

* delete CPUPlace and XPUPlace, test=develop

* enhance error messages of lookup_tale, merge_ids, data_norm

* fix

* fix error msg in .cu

* Support assignment to a Variable in dynamic mode. Note: not deal with backward.

* Rewrite VarBase __setitem__ for high-performance.

* try to test 3 means to do __setitem__ and test the performance of 3 means.

* Retain the means of the highest performance: C++ code and don't trace op.

* include ncclRecv and ncclSend, test=develop

* add gloo initializer, test=develop

* update name_scope example code

test=document_fix

* refine name_scope doc

test=document_fix

* fix example code

* rename var to tensor

* remove static

* fix return type

* modify sample code

* variable -> tensor

* migrate program_guard sample code

* refine error message

* migrate program_guard

* refine comment style

* fix indent

* add get final strategy for user to print final strategy