Skip to content

llama : make Qwen2MoE QKV bias optional #12477

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Mar 20, 2025
Merged

llama : make Qwen2MoE QKV bias optional #12477

merged 1 commit into from
Mar 20, 2025

Conversation

CISC
Copy link
Collaborator

@CISC CISC commented Mar 20, 2025

I'm guessing someone is cooking something, see huggingface/transformers#36735 :)

@CISC CISC requested a review from slaren March 20, 2025 10:57
@CISC
Copy link
Collaborator Author

CISC commented Mar 20, 2025

Hm, I see test-backend-ops fails a lot randomly lately, what's up with that?

@CISC CISC merged commit dbb3a47 into ggml-org:master Mar 20, 2025
47 of 48 checks passed
@CISC CISC deleted the qwen2moe_optional_qkv_bias branch March 20, 2025 11:50
@ggerganov
Copy link
Member

25: CPY(type_src=f32,type_dst=q5_1,ne=[256,2,3,4],permute=[0,2,1,3]): [CPY] NMSE = 0.000002013 > 0.000001000 FAIL

This failure is expected. I think it was caused by some discrepancy of how float numbers are rounded on the CPU and on Metal (if I remember correctly).

Ivy233 pushed a commit to Ivy233/llama.cpp that referenced this pull request Mar 23, 2025
@CISC @Ivy233
llama : make Qwen2MoE QKV bias optional (ggml-org#12477)
886e3c2
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@ggerganov ggerganov ggerganov approved these changes

@slaren slaren Awaiting requested review from slaren

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@CISC @ggerganov