]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
ggml webgpu: add support for soft_max, optimize rms_norm (llama/16357)
authorReese Levine <redacted>
Thu, 2 Oct 2025 18:00:31 +0000 (11:00 -0700)
committerGeorgi Gerganov <redacted>
Sun, 12 Oct 2025 04:57:25 +0000 (07:57 +0300)
commit4984916ab60314596493115e844dd7b28bee86cf
tree0d5316a14d3f5441abf239a24a8d66d465d29ef6
parent24ab95634d3969e81d2d5f1508d7b38628df0d2e
ggml webgpu: add support for soft_max, optimize rms_norm (llama/16357)

* Add inplace softmax

* Move rms_norm to split row approach

* Update debug for supports_op

* clean up debug statements

* Update tests/test-backend-ops.cpp

Co-authored-by: Georgi Gerganov <redacted>
---------

Co-authored-by: Georgi Gerganov <redacted>
include/ggml.h
src/ggml-webgpu/ggml-webgpu.cpp
src/ggml-webgpu/wgsl-shaders/rms_norm.wgsl
src/ggml-webgpu/wgsl-shaders/soft_max.tmpl.wgsl [new file with mode: 0644]
src/ggml.c
tests/test-backend-ops.cpp