ggml-webgpu: address precision issues for multimodal #22808
reeselevine
merged
13 commits into
ggml-org:master
from
noumena-labs:webgpu/fix-mul-max
fix(mixed-types): use f32 for precision and update the shared memory …
210766ce
fix(unary): correct the gelu, gelu quick and gelu erf functions
2380a062
fix(flash-attn-tile): fix the hardcode v type
f5f940fc
fix(flash_attn): fix tile path
f7b1560d
Merge branch 'master' of https://github.com/ggml-org/llama.cpp into w…
ef301b66
reeselevine
changed the title
Merge branch 'master' of https://github.com/ggml-org/llama.cpp into w…
2fb28310
fix: pass editorconfig and address the type conflicts
d2bd5ebd
fix: remove reduant pipeline keys
ab59cd5a
fix: remove inline min/max group size functions and revert the flash …
5a6f9c63
Merge branch 'master' of https://github.com/ggml-org/llama.cpp into w…
ffa63523
Merge branch 'master' of https://github.com/ggml-org/llama.cpp into w…
8c5597f2
fix: use clamp to avoid NaN for GELU
e115d544
fix: use the right range for exp, 80 is safer for f32 exp
6797d633
reeselevine
approved these changes
on 2026-05-12
CISC
approved these changes
on 2026-05-12
Reviewers
Assignees
No one assigned
Labels
ggml
WebGPU
Milestone
No milestone
Login to write a write a comment.
Login via GitHub