Lightning-AI/pytorch-lightning

enable loading `universal checkpointing` checkpoint in `DeepSpeedStrategy`

Open

#20,065 创建于 2024年7月9日

在 GitHub 查看
 (1 评论) (0 反应) (0 负责人)Python (26,687 star) (3,233 fork)batch import
featurehelp wantedstrategy: deepspeed

描述

Description & Motivation

After I trained a model in some numbers of gpus, say, 8 gpus for a while, It's difficult to load the checkpoint to 16 gpus with optimizer and model states unchanged. The deepspeed has developed the universal checkpointing strategy to solve this problem, but I didn't see the pytorch-lightning has this feature.

Pitch

I want the pytorch-lightning could support this feature

Alternatives

try to add universal_checkpoint as a param of DeepSpeedStrategy and modify the class refering to https://www.deepspeed.ai/tutorials/universal-checkpointing/

Additional context

No response

cc @borda @awaelchli

贡献者指南

技术栈
python
领域
backendmachine learning
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
5
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
over 1 week
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
fresh
清晰度议题是否清楚说明期望改动、验收标准和下一步。
mostly clear
前置要求
PyTorchDeepSpeedcheckpointing basics
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
10
研究方向
The issue requests adding universal checkpointing support to DeepSpeedStrategy. The contributor should study the DeepSpeed universal checkpointing tutorial at https://www.deepspeed.ai/tutorials/universal checkpointing/. The implementation likely involves modifying the DeepSpeedStrategy class in pytorch lightning's source code (e.g., lightning fabric/strategies/deepspeed.py). The maintainer comments from @borda and @awaelchli may provide guidance. The work includes adding a parameter to enable universal checkpoint and adapting the save/load methods.