* add readme in whl package

* fix chinese

* fix encoding

* del test

* upgrade gather_tree to core.ops (#30697)

* upgrade gather_tree to core.ops

* update gather_tree unittests

* update gather_tree doc (#30693)

* update gather_tree doc, test=document_fix

* update sample code, test=document_fix

* remove tensor type, test=document_fix

* fixed compilation error on gcc 4.8.x due to the usage of isfinite (#30733)

* modified conv+bn fuse pass to fix wrong mask in mask rcnn (#30704)

* Add cache for Transformer encoder.

* Bug fixed.

* add unittests for transformer encoder.

* add dla

* add python api
Co-authored-by: Nshangzhizhou <root@szth-rp-fanyi-opera49.szth.baidu.com>
Co-authored-by: Nshangzhizhou <root@szth-rp-fanyi-opera49.szth.baidu.com>

* Alias from  paddle.fluid.layers.auc to paddle.static.auc (#30206)

* add alias from  fluid.layers.auc to static.auc

* Update __init__.py

* add auc into all list

* alias acc, expose to users

* add auc into 'all' list (#30310)

* add auc into 'all' list

* alias acc, expose to users

* update sample code

* fix paddle.static.acc and auc sample code bug, test=document_fix

Co-authored-by: NJacek Czaja <jacek.czaja@intel.com>

* Add tf32 support for A100 tensor core acceleration for cuBLAS (#28732)

* Fixed an error

* Fixed an error

动态图中Conv2D保存成预测模型时，对应的Op可能是conv2d，也可能是depthwise_conv2d，但目前的save_quantized_model接口并未考虑depthwise_conv2d情况，可能会致使out_scale的值保存错误，该PR主要是修复这个问题。

* Fix the bug in fleet amp_init.

* Fix the amp_init unit test.

This PR is cherry-picked from PR: #29192
Function: Added TF32 switch for cuDNN. Turned on as default, turned off when users set the switch as False

[Dy2static]Fix paddle prefix in is_paddle_api (#30569)
cherry-pick #30569

* add fleet amp.init()

* add unittest for fleet_amp_init

* Implemented AddQuantDequantPass in imperative quantization.

* support 2.0 API such as Pool2D and ReLU

原始PR：#30485，#30507

* if pybind.cc changed, generate total report

Cherry pick PR #30520 .
Fix error message of Inplace strategy.

[cherry-pick] support layer_norm fp16 in dygraph amp (#30430)

cherry-pick #30553
fix bug of multicard grad ncclAllReduce, the gradient accumulater of parameters should be keep order, otherwsie, it will influence multicard ncclAllReduce of grad.

完善static.load的var_list参数。
当加载的是多个小文件时，Tensor列表可以是所有加载文件中Tensor的子集。
原始PR：#30457

cherry-pick #30536

Fix the compiling error of update_loss_scaling when using cuda9.

* add 2.0 API: device_guard

* add trainers for pserver

Change-Id: I99c0ab1cc427318f1f9bf8f8f5faff2b8890645d

* add trainers for pserver

Change-Id: I1a75793ec81ce126d07f4c47cae09b95d530bbc8

* support transformer v2.0

* fix range op crash in dygraph xpu place

* update voc url