Skip to content

Conversation

@lewtun
Copy link
Member

@lewtun lewtun commented May 21, 2025

We need this version to avoid OOM with Qwen3-8B on 1 node of 8 x H100s: deepspeedai/DeepSpeed#7258

SFT loss identical compared to v0.16.7:

Screenshot 2025-05-21 at 22 22 16

@lewtun lewtun requested a review from edbeeching May 21, 2025 20:08
@lewtun lewtun merged commit 8067149 into main May 21, 2025
1 check passed
@lewtun lewtun deleted the bump-deps branch May 21, 2025 20:25
lewtun added a commit that referenced this pull request May 21, 2025
Related to #653

(I forgot to include this in that PR)
lewtun added a commit that referenced this pull request May 22, 2025
Related to #653

(I forgot to include this in that PR)
mihailo417 added a commit to mihailo417/openR1 that referenced this pull request Oct 30, 2025
Related to huggingface/open-r1#653

(I forgot to include this in that PR)
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.

3 participants