fanqiNO1
GitHub
parent
3b639da1ef
commit
6c5eebb823
22
README.md
22
README.md
|
|
@ -58,29 +58,17 @@ English | [简体中文](README_zh-CN.md)
|
|||
|
||||
## What's New
|
||||
|
||||
v0.8.4 was released on 2023-08-03.
|
||||
v0.9.0 was released on 2023-10-10.
|
||||
|
||||
Highlights:
|
||||
|
||||
- Support enabling `efficient_conv_bn_eval` for efficient convolution and batch normalization. See [save memory on gpu](https://mmengine.readthedocs.io/en/latest/common_usage/save_gpu_memory.html#save-memory-on-gpu) for more details
|
||||
- Support training with [ColossalAI](https://colossalai.org/). Refer to the [Training Large Models](https://mmengine.readthedocs.io/en/latest/common_usage/large_model_training.html#colossalai) for more detailed usages.
|
||||
|
||||
- Add an [example](./examples/llama2/) to finetune Llama2.
|
||||
- Support gradient checkpointing. Refer to the [Save Memory on GPU](https://mmengine.readthedocs.io/en/latest/common_usage/save_gpu_memory.html#gradient-checkpointing) for more details.
|
||||
|
||||
- Support training with [FSDP](https://pytorch.org/tutorials/intermediate/FSDP_adavnced_tutorial.html?highlight=fsdp) and [DeepSpeed](https://www.deepspeed.ai/). Refer to the [Training Large Models](https://mmengine.readthedocs.io/en/latest/common_usage/large_model_training.html) for more detailed usages.
|
||||
- Supports multiple visualization backends, including `NeptuneVisBackend`, `DVCLiveVisBackend` and `AimVisBackend`. Refer to [Visualization Backends](https://mmengine.readthedocs.io/en/latest/common_usage/visualize_training_log.html) for more details.
|
||||
|
||||
- Introduce the pure Python style configuration file:
|
||||
|
||||
- Support navigating to base configuration file in IDE
|
||||
- Support navigating to base variable in IDE
|
||||
- Support navigating to source code of class in IDE
|
||||
- Support inheriting two configuration files containing the same field
|
||||
- Load the configuration file without other third-party requirements
|
||||
|
||||
Refer to the [tutorial](https://mmengine.readthedocs.io/en/latest/advanced_tutorials/config.html#a-pure-python-style-configuration-file-beta) for more detailed usages.
|
||||
|
||||
|
||||
|
||||
Read [Changelog](./docs/en/notes/changelog.md#v083-08032023) for more details.
|
||||
Read [Changelog](./docs/en/notes/changelog.md#v090-10102023) for more details.
|
||||
|
||||
## Table of Contents
|
||||
|
||||
|
|
|
|||
|
|
@ -58,29 +58,17 @@
|
|||
|
||||
## 最近进展
|
||||
|
||||
最新版本 v0.8.4 在 2023.08.03 发布。
|
||||
最新版本 v0.9.0 在 2023.10.10 发布。
|
||||
|
||||
亮点：
|
||||
|
||||
- 支持使用 `efficient_conv_bn_eval` 参数开启更高效的 `ConvBN` 推理模式。详见[节省显存文档](https://mmengine.readthedocs.io/zh_CN/latest/common_usage/save_gpu_memory.html)
|
||||
- 支持使用 [ColossalAI](https://colossalai.org/) 进行训练。可阅读[大模型训练](https://mmengine.readthedocs.io/zh_CN/latest/common_usage/large_model_training.html#colossalai)了解用法。
|
||||
|
||||
- 新增微调 Llama2 的[示例](./examples/llama2/)。
|
||||
- 支持梯度检查点。详见[用法](https://mmengine.readthedocs.io/zh_CN/latest/common_usage/save_gpu_memory.html#id3)。
|
||||
|
||||
- 支持使用 [FSDP](https://pytorch.org/tutorials/intermediate/FSDP_adavnced_tutorial.html?highlight=fsdp) 和 [DeepSpeed](https://www.deepspeed.ai/) 进行训练。可阅读[大模型训练](https://mmengine.readthedocs.io/zh_cn/latest/common_usage/large_model_training.html)了解用法。
|
||||
- 支持多种可视化后端，包括`NeptuneVisBackend`、`DVCLiveVisBackend` 和 `AimVisBackend`。可阅读[可视化后端](https://mmengine.readthedocs.io/zh_CN/latest/common_usage/visualize_training_log.html)了解用法。
|
||||
|
||||
- 引入纯 Python 风格的配置文件：
|
||||
|
||||
- 支持在 IDE 中导航到基础配置文件
|
||||
- 支持在 IDE 中导航到基础变量
|
||||
- 支持在 IDE 中导航到类的源代码
|
||||
- 支持继承包含相同字段的两个配置文件
|
||||
- 在加载配置文件时不需要其他第三方依赖
|
||||
|
||||
请参考[教程](https://mmengine.readthedocs.io/zh_CN/latest/advanced_tutorials/config.html#python-beta)以获取更详细的用法说明。
|
||||
|
||||
|
||||
|
||||
如果想了解更多版本更新细节和历史信息，请阅读[更新日志](./docs/en/notes/changelog.md#v083-08032023)
|
||||
如果想了解更多版本更新细节和历史信息，请阅读[更新日志](./docs/en/notes/changelog.md#v090-10102023)
|
||||
|
||||
## 目录
|
||||
|
||||
|
|
|
|||
|
|
@ -1,5 +1,69 @@
|
|||
# Changelog of v0.x
|
||||
|
||||
## v0.9.0 (10/10/2023)
|
||||
|
||||
### Highlights
|
||||
|
||||
- Support training with [ColossalAI](https://colossalai.org/). Refer to the [Training Large Models](https://mmengine.readthedocs.io/en/latest/common_usage/large_model_training.html#colossalai) for more detailed usages.
|
||||
- Support gradient checkpointing. Refer to the [Save Memory on GPU](https://mmengine.readthedocs.io/en/latest/common_usage/save_gpu_memory.html#gradient-checkpointing) for more details.
|
||||
- Supports multiple visualization backends, including `NeptuneVisBackend`, `DVCLiveVisBackend` and `AimVisBackend`. Refer to [Visualization Backends](https://mmengine.readthedocs.io/en/latest/common_usage/visualize_training_log.html) for more details.
|
||||
|
||||
### New Features & Enhancements
|
||||
|
||||
- Add a text translation example by [@Desjajja](https://github.com/Desjajja) in https://github.com/open-mmlab/mmengine/pull/1283
|
||||
- Add `NeptuneVisBackend` by [@wangerlie](https://github.com/wangerlie) in https://github.com/open-mmlab/mmengine/pull/1311
|
||||
- Add ColossalAI strategy by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1299
|
||||
- Add collect_results support for Ascend NPU by [@xuuyangg](https://github.com/xuuyangg) in https://github.com/open-mmlab/mmengine/pull/1309
|
||||
- Unify the parameter style of DeepSpeedStrategy by [@LZHgrla](https://github.com/LZHgrla) in https://github.com/open-mmlab/mmengine/pull/1320
|
||||
- Add progressbar rich by [@Dominic23331](https://github.com/Dominic23331) in https://github.com/open-mmlab/mmengine/pull/1157
|
||||
- Support using other file handlers by [@KevinNuNu](https://github.com/KevinNuNu) in https://github.com/open-mmlab/mmengine/pull/1188
|
||||
- Refine error message by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1338
|
||||
- Implement gradient checkpointing by [@zeyuanyin](https://github.com/zeyuanyin) in https://github.com/open-mmlab/mmengine/pull/1319
|
||||
- Add `DVCLiveVisBackend` by [@RangeKing](https://github.com/RangeKing) in https://github.com/open-mmlab/mmengine/pull/1336
|
||||
- Add `AimVisBackend` by [@RangeKing](https://github.com/RangeKing) in https://github.com/open-mmlab/mmengine/pull/1347
|
||||
- Support bitsandbytes by [@okotaku](https://github.com/okotaku) in https://github.com/open-mmlab/mmengine/pull/1357
|
||||
- Support `Adafactor` Optimizer by [@okotaku](https://github.com/okotaku) in https://github.com/open-mmlab/mmengine/pull/1361
|
||||
- Add unit tests for autocast with Ascend device by [@6Vvv](https://github.com/6Vvv) in https://github.com/open-mmlab/mmengine/pull/1363
|
||||
- Support metainfo of dataset can be a generic dict-like Mapping by [@hiyyg](https://github.com/hiyyg) in https://github.com/open-mmlab/mmengine/pull/1378
|
||||
- Support for installing minimal runtime dependencies by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1362
|
||||
- Support setting the number of iterations in `Runner` for each epoch by [@ShuRaymond](https://github.com/ShuRaymond) in https://github.com/open-mmlab/mmengine/pull/1292
|
||||
- Support using gradient checkpointing in FSDP by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1382
|
||||
|
||||
### Docs
|
||||
|
||||
- Add README for examples by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1295
|
||||
- Add a new ecosystem in README by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1296
|
||||
- Fix typo by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1298
|
||||
- Add an image for Neptune by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1312
|
||||
- Fix docs of ColossalAI by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1315
|
||||
- Update QRCode by [@crazysteeaam](https://github.com/crazysteeaam) in https://github.com/open-mmlab/mmengine/pull/1328
|
||||
- Add activation checkpointing usage by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1341
|
||||
- Fix typo by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1348
|
||||
- Update the usage of bitsandbytes in Chinese documents by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1359
|
||||
- Fix doc typo our_dir in LoggerHook by [@wangg12](https://github.com/wangg12) in https://github.com/open-mmlab/mmengine/pull/1373
|
||||
- Add the contributing doc in pr template by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1380
|
||||
- Update the version info by [@fanqiNO1](https://github.com/fanqiNO1) in https://github.com/open-mmlab/mmengine/pull/1383
|
||||
- Fix typo by [@fanqiNO1](https://github.com/fanqiNO1) in https://github.com/open-mmlab/mmengine/pull/1385
|
||||
|
||||
### Bug Fixes
|
||||
|
||||
- Ignore examples in CI by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1297
|
||||
- Fix multi-card issue in PyTorch v2.1 on Ascend by [@LRJKD](https://github.com/LRJKD) in https://github.com/open-mmlab/mmengine/pull/1321
|
||||
- Fix get `optimizer_cls` by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1324
|
||||
- Fix ndarray metainfo check in ConcatDataset by [@NrealLzx](https://github.com/NrealLzx) in https://github.com/open-mmlab/mmengine/pull/1333
|
||||
- Adapt to PyTorch v2.1 on Ascend by [@LRJKD](https://github.com/LRJKD) in https://github.com/open-mmlab/mmengine/pull/1332
|
||||
- Fix the type check of tasks in progress bar by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1340
|
||||
- The keyword mode appears nested multiple times in the log by [@huaibovip](https://github.com/huaibovip) in https://github.com/open-mmlab/mmengine/pull/1305
|
||||
- Fix pydantic version to fix mlflow unit tests by [@zhouzaida](https://github.com/zhouzaida) in https://github.com/open-mmlab/mmengine/pull/1351
|
||||
- Fix get class attribute from a string by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1345
|
||||
- Delete yapf verify by [@okotaku](https://github.com/okotaku) in https://github.com/open-mmlab/mmengine/pull/1365
|
||||
- Ensure from_cfg of Runner have the same defaults values as its __init__ by [@YinAoXiong](https://github.com/YinAoXiong) in https://github.com/open-mmlab/mmengine/pull/1368
|
||||
- Fix docs building error caused by deepspeed by [@HAOCHENYE](https://github.com/HAOCHENYE) in https://github.com/open-mmlab/mmengine/pull/1379
|
||||
|
||||
### Contributors
|
||||
|
||||
A total of 21 developers contributed to this release. Thanks [@LZHgrla](https://github.com/LZHgrla), [@wangerlie](https://github.com/wangerlie), [@wangg12](https://github.com/wangg12), [@RangeKing](https://github.com/RangeKing), [@hiyyg](https://github.com/hiyyg), [@LRJKD](https://github.com/LRJKD), [@KevinNuNu](https://github.com/KevinNuNu), [@zeyuanyin](https://github.com/zeyuanyin), [@Desjajja](https://github.com/Desjajja), [@ShuRaymond](https://github.com/ShuRaymond), [@okotaku](https://github.com/okotaku), [@crazysteeaam](https://github.com/crazysteeaam), [@6Vvv](https://github.com/6Vvv), [@NrealLzx](https://github.com/NrealLzx), [@YinAoXiong](https://github.com/YinAoXiong), [@huaibovip](https://github.com/huaibovip), [@xuuyangg](https://github.com/xuuyangg), [@Dominic23331](https://github.com/Dominic23331), [@fanqiNO1](https://github.com/fanqiNO1), [@HAOCHENYE](https://github.com/HAOCHENYE), [@zhouzaida](https://github.com/zhouzaida)
|
||||
|
||||
## v0.8.4 (03/08/2023)
|
||||
|
||||
### New Features & Enhancements
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
# Copyright (c) OpenMMLab. All rights reserved.
|
||||
|
||||
__version__ = '0.8.4'
|
||||
__version__ = '0.9.0'
|
||||
|
||||
|
||||
def parse_version_info(version_str):
|
||||
|
|
|
|||
Loading…
Reference in New Issue