Skip to content

convert : force patch_merger tensors to f16/f32#18124

Merged
CISC merged 1 commit intomasterfrom
cisc/convert-force-patch-merger-f16-f32
Dec 17, 2025
Merged

convert : force patch_merger tensors to f16/f32#18124
CISC merged 1 commit intomasterfrom
cisc/convert-force-patch-merger-f16-f32

Conversation

@CISC
Copy link
Member

@CISC CISC commented Dec 17, 2025

Same fix as in #15367

Fixes #18122

@ngxson
Copy link
Contributor

ngxson commented Dec 17, 2025

probably need to update the list of skipped tensors in llama-quant.cpp too

@CISC
Copy link
Member Author

CISC commented Dec 17, 2025

probably need to update the list of skipped tensors in llama-quant.cpp too

Hmmm, right, looks like none of these are there...

@CISC CISC merged commit 5166aaf into master Dec 17, 2025
9 checks passed
@CISC CISC deleted the cisc/convert-force-patch-merger-f16-f32 branch December 17, 2025 21:15
Anico2 added a commit to Anico2/llama.cpp that referenced this pull request Jan 15, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

python python script changes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Eval bug: [CUDA] GLM-4.6V-Flash crashes when using BF16 mmproj

2 participants