InternEvo
InternEvo copied to clipboard
InternEvo is an open-sourced lightweight training framework aims to support model pre-training without the need for extensive dependencies.
### 描述该功能 目前仓库对pip安装的依赖库版本有严格的指定,对于广泛的使用会有一定的不便。需要逐步放松对应的限制 ### 是否希望自己实现该功能? - [ ] 我希望自己来实现这一功能,并向 InternLM 贡献代码!
### Describe the feature Current implementation does not support partially training, i.e., part of the model parameters are frozen. If I understand it correctly, the assertion around line 584 in...
### Describe the feature 在目前的情形下,pipeline parallel会在第一份数据进入时固定logits等的形状,但事实上,下游任务中很多时候最好能支持变长以节省算力,而不是进行过多的padding。 ### Will you implement it? - [ ] I would like to implement this feature and create a PR!
Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...
## Motivation - **support pip install**: one-stop installation via python packaging - **easier API**: from high-level 'TrainerBuilder', to mid-level ZeRO/FSDP/TP/PP/WP/SP/CP/EP parallelism, and low-level ops - **FSDP+EP**: support FSDP working with...
### Describe the feature 如题 ### Will you implement it? - [ ] I would like to implement this feature and create a PR!
Test for SC++ implements on NPU, bug fixes
Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...
add non p2p communication for pipeline scheduling
Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...