llama.cpp
gemma : perform per-layer projections in the first layer
#21612
Merged

gemma : perform per-layer projections in the first layer #21612

ggerganov merged 3 commits into master from gg/models-per-layer-fixes-2
ggerganov
ggerganov ggerganov requested a review from CISC CISC 7 days ago
ggerganov ggerganov requested a review from ngxson ngxson 7 days ago
ngxson
ngxson approved these changes on 2026-04-08
ggerganov gemma : reduce graph splits by keeping per-layer ops in the input layer
9787dd2b
ggerganov gemma : put the per-layer proj in the first layer
c8752f83
ggerganov cont : move the projection before the layer loop
f4663616
ggerganov ggerganov force pushed from 3b29d142 to f4663616 7 days ago
ggerganov
ngxson
ngxson approved these changes on 2026-04-08
ggerganov ggerganov merged 5764d7c6 into master 7 days ago
ggerganov ggerganov deleted the gg/models-per-layer-fixes-2 branch 7 days ago
taronaeo

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone