Skip to content

Fix enable_model_cpu_offload problems#320

Merged
a-r-r-o-w merged 3 commits into
mainfrom
fix/enable-model-cpu-offload
Mar 14, 2025
Merged

Fix enable_model_cpu_offload problems#320
a-r-r-o-w merged 3 commits into
mainfrom
fix/enable-model-cpu-offload

Conversation

@a-r-r-o-w
Copy link
Copy Markdown
Contributor

Fixes #212, #295

cc @dorpxam Could you give this a try? I believe it should fix any issues you're facing when using enable_model_cpu_offload

This was referenced Mar 13, 2025
@dorpxam
Copy link
Copy Markdown

dorpxam commented Mar 14, 2025

Just see that one. For me precomputation is not more an issue with latest updates. Thanks so much!

@a-r-r-o-w a-r-r-o-w merged commit b8bf0fc into main Mar 14, 2025
@a-r-r-o-w a-r-r-o-w deleted the fix/enable-model-cpu-offload branch March 14, 2025 21:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

enable_model_cpu_offload causes NCCL timeout during multi-gpu training

2 participants