]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
CUDA: fix FA FP16 accumulator overflow for Granite (llama/18614)
authorJohannes Gäßler <redacted>
Mon, 5 Jan 2026 18:51:13 +0000 (19:51 +0100)
committerGeorgi Gerganov <redacted>
Sun, 11 Jan 2026 09:02:08 +0000 (11:02 +0200)
commitbb7e7d2dd98763a34442b524218ead005d08d2fa
treea0dd88661801b2eee6aa6f1a70d4e3019e005e34
parente48d756595635fa947806b730b32e45b5a6ef2cd
CUDA: fix FA FP16 accumulator overflow for Granite (llama/18614)
src/ggml-cuda/fattn-common.cuh