Update the code for the min,max, test=develop

add the api and doc for the margin_ranking_loss and MarginRankingLoss

* move paddle.fleet to paddle.distributed.fleet

* add auto_cast, test=develop

* add loss scaler, test=develop

* add comments, test=develop

* refine code, test=develop

* refine code, test=develop

* do not set flags automatically, test=develop

* fix custom op bug, test=develop

* add more test, test=develop

* refine enable logic, test=develop

* enable amp test with GPU, test=develop

* add unittest

* add test for found_inf

* follow comments

* follow comments

* remove global variable, use singleton

* add some notes

* update comments

* update comments

* update comments

* add use_dynamic_loss_scaling argument

* refine found_inf

* refine found_inf

* fix loaded var suffix repeat error

* use new dygraph name for loaded param

change api name eps to epsilon for the pair_distance 

* add nll loss API, update demo code of the comment

[Dy2Stat] Removes temporary files created during the transformation of dygraph to static graph.  (#26150)

add pairewise distance for the paddlepaddle api 2.0

fix paddle.nn.loss.L1Loss OP, add paddle.nn.functional.l1_loss OP for API2.0

* Add the max, min, maximum, minimum api for the API 2.0, test=develop

* fixed imperative module in doc example code

* fixed static module

* solve conflict

* add IterableDataset support in multiprocess DataLoader. test=develop

add lamb to fleet meta optimizer

add deprecated decorator, remove alias for cumsum op

* - introducing oneDNN 1.6

test=develop

* - onednn 1.5

test=develop

* - oneDNN 1.5

test=develop

* - oneDNN 1.5 bugfix

test=develop

* Directory migration, test=develop

* Change imperative from paddle init to paddle framework, test=develop

* Fixed jit bug, test=develop

* default static mode, test=develop

* fixed format and create parameter belongs to framework, test=develop

* Fixed import package, test=develop

* fix __init__ format, test=develop

* fixed alias problem

* fixed paddle.enable_imperative problems, test=develop

* Add unittest

* delete install_check comment

* Fixed unittest timeout

* fixed unittest error

* move Program default_xx_program to static package

* optimize unittest method

* fixed framework __init__ format

* fixed jit path

* delete alias

* move jit to paddle

* Fixed unittest format

* fixed paddle.default_main_program

* Fixed save load API in paddle __init__.py

* fixed ci paddle.imperative.to_variable

* add the support for multi-node training

update cumsum api and fix up the cumsum op

* fix prelu for compatibility with saved model of old version

* reshape alpha

* Add oneDNN relu6 op

* Lint fixes

* fix merge3 test=develop

* add paddle.fleet.AsyncOptimizer
Co-authored-by: Ndongdaxiang <dongdaxiang@baidu.com>

* add pin memory control

* fix buffered reader init problem

* fix unittest error

* add unittest for coverage

* add cuda pinned place branch

* add unittest

* add skip when not gpu

* support multi-ps training mode for fleetrun; test=develop

1. update document template for unary elementwise layers(a.k.a. activation layer);
2. remove generate_op_noattr and use generate_activation instead; remove redundant function copies;
3. minor update for docstring to fix rst format errors.
4. fix doc for Rsqrt OP
5. add sample code for each activation separately;
6. remove the unused deprecated decorator.

* Change use_quantizer attribute name and data type

* Fix problem with setting attribute

* Add changes due to review

* Small change in function

* Restore use_quantizer attr for compatibility

* enable check_dygraph for mkldnn activation tests

* TestMKLDNNSwishDim4 float32->self.dtype

* refine unsqueeze, test=develop

* update unsqueeze, test=develop

* refine unsqueeze, test=develop

* refine unsqueeze, test=develop

* update

* remove None, test=develop

* follow comments

* support bool

* update doc

* follow comments

* merge develop