llama.cpp
models: fix GDN normalization from `max` to `rsqrt`
#28068

Merged

models: fix GDN normalization from `max` to `rsqrt` #28068

CISC merged 2 commits into ggml-org:master from danielhanchen:gdn/fla-l2norm

models: use flash-linear-attention's l2norm for gated delta net q/k

c03dc68b

github-actions added model

danielhanchen marked this pull request as ready for review 13 days ago

danielhanchen requested a review from

CISC 13 days ago

ggerganov commented on 2026-08-31

Update src/models/models.h

757abc99

CISC approved these changes on 2026-08-31

ggerganov approved these changes on 2026-08-31

CISC merged 5fdfa628 into master 7 days ago

Reviewers

ggerganov

CISC

Assignees

No one assigned

Labels

model

Milestone

No milestone