Skip to content

Commit

Permalink
update release note (PaddlePaddle#3719) (PaddlePaddle#3720)
Browse files Browse the repository at this point in the history
  • Loading branch information
@TCChenlong
TCChenlong authored Aug 12, 2021
1 parent 950d231 commit 49c48f6
Show file tree
Hide file tree
Showing 2 changed files with 50 additions and 53 deletions.
52 changes: 25 additions & 27 deletions docs/release_note_cn.md
View file
Original file line number Diff line number Diff line change
@@ -1,68 +1,66 @@


# 2.1.2 Release Note

## 重要更新

本版本主要是对2.1.1中一些功能和性能问题的修复，重点如下：

- 修复了动转静语法转写已知的若干问题
- 自定义OP编译时C++版本检查由C++11升级为C++14
- 修复了基础API中的已知问题。
- 修复了动转静语法转写已知的若干问题。
- 自定义OP编译时C++版本检查由C++11升级为C++14。

## 训练框架

### 功能优化（含分布式）

#### 基础API

- 恢复 `paddle.vision.xxx` 路径模块可调用性，增强对旧版本的兼容；但是不推荐使用该路径调用API([#34489](https://github.com/PaddlePaddle/Paddle/pull/34489))
- 修复`paddle.concat`在应用到多个大`shape` 的Tensor时溢出的问题([#34396](https://github.com/PaddlePaddle/Paddle/pull/34396))
- `paddle.flip`支持输入axis为整型，并提升了动态图模式下的性能([#34477](https://github.com/PaddlePaddle/Paddle/pull/34477))
- 修复`paddle.slice` 输入输出地址相同时越界访问问题([#34265](https://github.com/PaddlePaddle/Paddle/pull/34265))
- 修复 `paddle.nn.Unfold` 的输入参数顺序错误的问题([#34251](https://github.com/PaddlePaddle/Paddle/pull/34251))
- 新增了静态图下 `Tensor` 的若干接口，如 `size()、detach()`等 ([#33330](https://github.com/PaddlePaddle/Paddle/pull/33330))
- `Tensor.grad` 的 Warning内容中增加了不兼容升级的说明 ([#34262](https://github.com/PaddlePaddle/Paddle/pull/34264))
- 下线 `paddle.save` 保存 `Layer` 的功能([#34039](https://github.com/PaddlePaddle/Paddle/pull/34039))
- 修复 `paddle.jit.save`在Mac系统上保存的模型，在Linux平台上无法对模型进行重训练的问题([#34154](https://github.com/PaddlePaddle/Paddle/pull/34154))
- 修复 `layer_norm` 在大 `size` 输入时 `cuda kernel` 参数错误的问题([#33893](https://github.com/PaddlePaddle/Paddle/pull/33893))
- 修复`paddle.io.DataLoader`误报不兼容升级warning问提([#34001](https://github.com/PaddlePaddle/Paddle/pull/34001))
- 修复`paddle.io.DataLoader`内存泄漏问题([#34301](https://github.com/PaddlePaddle/Paddle/pull/34301))
- 修复 `paddle.vision` 路径下部分API无法访问的问题。([#34489](https://github.com/PaddlePaddle/Paddle/pull/34489))
- 修复`paddle.concat`在应用到多个大`shape` 的Tensor时溢出的问题。([#34396](https://github.com/PaddlePaddle/Paddle/pull/34396))
- `paddle.flip`支持输入axis为整型，并提升了动态图模式下的性能。([#34477](https://github.com/PaddlePaddle/Paddle/pull/34477))
- 修复`paddle.slice` 输入输出地址相同时越界访问问题。([#34265](https://github.com/PaddlePaddle/Paddle/pull/34265))
- 修复 `paddle.nn.Unfold` 的输入参数顺序错误的问题。([#34251](https://github.com/PaddlePaddle/Paddle/pull/34251))
- 新增了静态图下 `Tensor` 的若干接口，如 `size()、detach()`等。 ([#33330](https://github.com/PaddlePaddle/Paddle/pull/33330))
- `Tensor.grad` 的 Warning内容中增加了不兼容升级的说明。([#34262](https://github.com/PaddlePaddle/Paddle/pull/34264))
- 下线 `paddle.save` 保存 `Layer` 的功能。([#34039](https://github.com/PaddlePaddle/Paddle/pull/34039))
- 修复 `paddle.jit.save`在Mac系统上保存的模型，在Linux平台上无法对模型进行重训练的问题。([#34154](https://github.com/PaddlePaddle/Paddle/pull/34154))
- 修复 `layer_norm` 在大 `size` 输入时 `cuda kernel` 参数错误的问题。([#33893](https://github.com/PaddlePaddle/Paddle/pull/33893))
- 修复`paddle.io.DataLoader`误报不兼容升级warning问题。([#34001](https://github.com/PaddlePaddle/Paddle/pull/34001))
- 修复`paddle.io.DataLoader`内存泄漏问题。([#34301](https://github.com/PaddlePaddle/Paddle/pull/34301))

#### 动态图转静态图

- 新增对 `Sequential` 容器类嵌套使用时的语法支持([#34246](https://github.com/PaddlePaddle/Paddle/pull/34262))
- 新增对 `Python3 type hint` 语法的兼容支持([#33745](https://github.com/PaddlePaddle/Paddle/pull/33745))
- `@to_static` 中 `input_spec `参数新增支持非 `Tensor` 类型，如 `int、float、string、bool`等([#33464](https://github.com/PaddlePaddle/Paddle/pull/33464))
- 修复了动转静语法转写已知的若干问题([#33963](https://github.com/PaddlePaddle/Paddle/pull/33963))
- 新增对 `Sequential` 容器类嵌套使用时的语法支持。([#34246](https://github.com/PaddlePaddle/Paddle/pull/34262))
- 新增对 `Python3 type hint` 语法的兼容支持。([#33745](https://github.com/PaddlePaddle/Paddle/pull/33745))
- `@to_static` 中 `input_spec `参数新增支持非 `Tensor` 类型，如 `int、float、string、bool`等。([#33464](https://github.com/PaddlePaddle/Paddle/pull/33464))
- 修复了动转静语法转写已知的若干问题。([#33963](https://github.com/PaddlePaddle/Paddle/pull/33963))

#### 自定义OP

- 自定义OP编译时C++版本检查由C++11升级为C++14 ([#30415](https://github.com/PaddlePaddle/Paddle/pull/34015))

- 自定义OP编译时C++版本检查由C++11升级为C++14。 ([#30415](https://github.com/PaddlePaddle/Paddle/pull/34015))

## 推理部署


### Paddle Inference

#### 问题修复

- 修复`batch_size > 1`时ERNIE模型计算结果错误的问题([#33784](https://github.com/PaddlePaddle/Paddle/pull/33784))
- 修复`batch_size > 1`时ERNIE模型计算结果错误的问题。([#33784](https://github.com/PaddlePaddle/Paddle/pull/33784))
- 修复windows下`TensortRT`推理路径用右斜杠分割导致的崩溃。([#33885](https://github.com/PaddlePaddle/Paddle/pull/33885))
- 修复MKLDNN `elementwise`系列OP的X不支持广播的问题（[#33845](https://github.com/PaddlePaddle/Paddle/pull/33845)）
- 修复MKLDNN `elementwise`系列OP的X不支持广播的问题。（[#33845](https://github.com/PaddlePaddle/Paddle/pull/33845)）

## 环境适配

### 编译安装

- 限定了依赖的 Gast 库的版本范围( `gast>=0.3.3, <=0.4.0`)([#33850](https://github.com/PaddlePaddle/Paddle/pull/33850))
- 优化了`Avx/No-Avx`相关的安装报错信息，减少了冗余的Warning信息([#33885](https://github.com/PaddlePaddle/Paddle/pull/33905))
- 限定了依赖的 Gast 库的版本范围。( `gast>=0.3.3, <=0.4.0`)([#33850](https://github.com/PaddlePaddle/Paddle/pull/33850))
- 优化了`Avx/No-Avx`相关的安装报错信息，减少了冗余的Warning信息。([#33885](https://github.com/PaddlePaddle/Paddle/pull/33905))

### 新硬件适配

#### 昆仑硬件训练支持

- 修改昆仑的`cmake`文件，统一更新昆仑的算子库（[#34000](https://github.com/PaddlePaddle/Paddle/pull/34000)）
- 修改昆仑的`cmake`文件，统一更新昆仑的算子库。（[#34000](https://github.com/PaddlePaddle/Paddle/pull/34000)）

## Thanks to our Contributors

Expand Down
51 changes: 25 additions & 26 deletions docs/release_note_en.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,63 +6,63 @@

This release mainly fixes some features and performance issues in 2.1.1. See the following highlights:

- Fix several known problems with dynamic to static syntax transcriptions
- C++ version check upgrade from C++11 to C++14 during Custom OP compile
- Fix several known issues in frontend APIs.
- Fix several known issues with dynamic to static syntax transcriptions.
- C++ version check upgrade from C++11 to C++14 during Custom OP compile.

## Training framework

### Functional optimization (including distributed)

#### Basic API

- Restore the callability of the `paddle.vision.xxx` path module to enhance compatibility of the old versions; yet it is not recommended to use this path to call the API([#34489](https://github.com/PaddlePaddle/Paddle/pull/34489))
- Fix `paddle.concat` overflow when applied to multiple Tensor with large `shape` ([#34396](https://github.com/PaddlePaddle/Paddle/pull/34396))
- `paddle.flip` supports input axis as integer, and improves performance in dynamic graph mode ([#34477](https://github.com/PaddlePaddle/Paddle/pull/34477))
- Fix `paddle.slice` out-of-bounds access problem when input and output addresses are the same([#34265](https://github.com/PaddlePaddle/Paddle/pull/34265))
- Fix the problem of wrong order of input parameters of `paddle.nn.Unfold`([#34251](https://github.com/PaddlePaddle/Paddle/pull/34251))
- Fix some APIs located in `paddle.vision` are not accessible issues. ([#34489](https://github.com/PaddlePaddle/Paddle/pull/34489))
- Fix `paddle.concat` overflow when applied to multiple Tensor with large `shape`. ([#34396](https://github.com/PaddlePaddle/Paddle/pull/34396))
- `paddle.flip` supports input axis as integer, and improves performance in dynamic graph mode. ([#34477](https://github.com/PaddlePaddle/Paddle/pull/34477))
- Fix `paddle.slice` out-of-bounds access problem when input and output addresses are the same. ([#34265](https://github.com/PaddlePaddle/Paddle/pull/34265))
- Fix the problem of wrong order of input parameters of `paddle.nn.Unfold`. ([#34251](https://github.com/PaddlePaddle/Paddle/pull/34251))
- Add several interfaces for `Tensor` under static graphs such as `size(), detach()`, etc. ([#33330](https://github.com/PaddlePaddle/Paddle/pull/33330))
- Add incompatible upgrade note to the Warning content of `Tensor.grad`([#34262](https://github.com/PaddlePaddle/Paddle/pull/34264))
- Downlink `paddle.save` to save the function of `Layer`([#34039](https://github.com/PaddlePaddle/Paddle/pull/34039))
- Fix `paddle.jit.save` for saving models on Mac systems that cannot be retrained on Linux platforms([#34154](https://github.com/PaddlePaddle/Paddle/pull/34154))
- Fix `layer_norm` with wrong `cuda kernel` parameters for large `size` input ([#33893](https://github.com/PaddlePaddle/Paddle/pull/33893))
- Fix `paddle.io.DataLoader` error reporting incompatible upgrade warning issue ([#34001](https://github.com/PaddlePaddle/Paddle/pull/34001))
- Fix `paddle.io.DataLoader` memory leak problem([#34301](https://github.com/PaddlePaddle/Paddle/pull/34301))
- Add incompatible upgrade note to the Warning content of `Tensor.grad`.([#34262](https://github.com/PaddlePaddle/Paddle/pull/34264))
- Downlink `paddle.save` to save the function of `Layer`. ([#34039](https://github.com/PaddlePaddle/Paddle/pull/34039))
- Fix `paddle.jit.save` for saving models on Mac systems that cannot be retrained on Linux platforms. ([#34154](https://github.com/PaddlePaddle/Paddle/pull/34154))
- Fix `layer_norm` with wrong `cuda kernel` parameters for large `size` input. ([#33893](https://github.com/PaddlePaddle/Paddle/pull/33893))
- Fix `paddle.io.DataLoader` error reporting incompatible upgrade warning issue. ([#34001](https://github.com/PaddlePaddle/Paddle/pull/34001))
- Fix `paddle.io.DataLoader` memory leak problem. ([#34301](https://github.com/PaddlePaddle/Paddle/pull/34301))

#### Dynamic to static map

- Add syntax support for nested use of `Sequential` container classes ([#34246](https://github.com/PaddlePaddle/Paddle/pull/34262))
- Add compatibility support for `Python3 type hint` syntax([#33745](https://github.com/PaddlePaddle/Paddle/pull/33745))
- Add support for non-`Tensor` types including `int, float, string, bool` in the `input_spec` argument of `@to_static`([#33464](https://github.com/PaddlePaddle/Paddle/pull/33464))
- Fix a number of known problems with the transcription of dynamic to static syntax ([#33963](https://github.com/PaddlePaddle/Paddle/pull/33963))
- Add syntax support for nested use of `Sequential` container classes. ([#34246](https://github.com/PaddlePaddle/Paddle/pull/34262))
- Add compatibility support for `Python3 type hint` syntax. ([#33745](https://github.com/PaddlePaddle/Paddle/pull/33745))
- Add support for non-`Tensor` types including `int, float, string, bool` in the `input_spec` argument of `@to_static`. ([#33464](https://github.com/PaddlePaddle/Paddle/pull/33464))
- Fix a number of known problems with the transcription of dynamic to static syntax. ([#33963](https://github.com/PaddlePaddle/Paddle/pull/33963))

#### Custom OP

- C++ version check upgrade from C++11 to C++14 during Custom OP compile ([#30415](https://github.com/PaddlePaddle/Paddle/pull/34015))
- C++ version check upgrade from C++11 to C++14 during Custom OP compile. ([#30415](https://github.com/PaddlePaddle/Paddle/pull/34015))


## Inference Deployment


### Paddle Inference

#### Issue fix
#### bugfix

- Fix wrong calculation result of ERNIE model when `batch_size > 1` ([#33784](https://github.com/PaddlePaddle/Paddle/pull/33784))
- Fix wrong calculation result of ERNIE model when `batch_size > 1`. ([#33784](https://github.com/PaddlePaddle/Paddle/pull/33784))
- Fix the crash caused by splitting `TensortRT` inference path with right slash under windows.([#33885](https://github.com/PaddlePaddle/Paddle/pull/33885))
- Fix MKLDNN `elementwise` series OP's X does not support broadcast （[#33845](https://github.com/PaddlePaddle/Paddle/pull/33845)）
- Fix MKLDNN `elementwise` series OP's X does not support broadcast .（[#33845](https://github.com/PaddlePaddle/Paddle/pull/33845)）

## Environment adaptation

### Compile and install

- Restrict the version range of dependent Gast libraries ( `gast>=0.3.3, <=0.4.0`)([#33850](https://github.com/PaddlePaddle/Paddle/pull/33850))
- Optimize `Avx/No-Avx` related installation error messages, reduce redundant Warning messages([#33885](https://github.com/PaddlePaddle/Paddle/pull/33905))
- Restrict the version range of dependent Gast libraries ( `gast>=0.3.3, <=0.4.0`). ([#33850](https://github.com/PaddlePaddle/Paddle/pull/33850))
- Optimize `Avx/No-Avx` related installation error messages, reduce redundant Warning messages. ([#33885](https://github.com/PaddlePaddle/Paddle/pull/33905))

### New Hardware Adaptation

#### Kunlun hardware training support

- Modify the `cmake` file of Kunlun to unify and update its operator library（[#34000](https://github.com/PaddlePaddle/Paddle/pull/34000)）
- Modify the `cmake` file of Kunlun to unify and update its operator library.（[#34000](https://github.com/PaddlePaddle/Paddle/pull/34000)）

## Thanks to our Contributors

Expand All @@ -71,7 +71,6 @@ This release contains contributions from:
0x45f、Aurelius84、Chen Weihang、chentianyu03、HexToString、iducn、Jacek Czaja、Kaipeng Deng、Leo Chen、lzzyzlbb、Peihan、taixiurong、tianshuo78520a、WeiXin、wenbin、Wilber、wuhuachaocoding、xiongkun、Zhou Wei、 winter-wang .



# 2.1.1 Release Note

## Important Updates
Expand Down

0 comments on commit 49c48f6

Please sign in to comment.