]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
CANN: support gated linear attn (llama/18653)
authorhipudding <redacted>
Fri, 16 Jan 2026 08:18:49 +0000 (16:18 +0800)
committerGeorgi Gerganov <redacted>
Fri, 30 Jan 2026 11:49:29 +0000 (13:49 +0200)
commitaab6aace29c58362c15185ff9f280e72db4302aa
tree0a254acd5e5ff356bcca5bf4705bc45aea0de8c2
parent1999fc0a82637bb9632a1c11957da948031e9b44
CANN: support gated linear attn (llama/18653)

* CANN: support gated linear attn

This change adds support for the GGML_OP_GATED_LINEAR_ATTN operator.
The feature was implemented by YushengZhao. Because the previous
submission was based on an outdated codebase, this PR was rebased to
merge.

Co-authored-by: YushengZhao <redacted>
Co-authored-by: hipudding <redacted>
* CANN: optimize OP gla

Optimize gla for high preformance

* Remove unused comments

---------

Co-authored-by: 赵禹昇 <redacted>
Co-authored-by: YushengZhao <redacted>
src/ggml-cann/aclnn_ops.cpp
src/ggml-cann/aclnn_ops.h
src/ggml-cann/ggml-cann.cpp