]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
CUDA: fix FP16 overflow in tile FA kernel (llama/17875)
authorJohannes Gäßler <redacted>
Tue, 9 Dec 2025 08:34:02 +0000 (09:34 +0100)
committerGeorgi Gerganov <redacted>
Thu, 11 Dec 2025 13:32:59 +0000 (15:32 +0200)
commit16e3125bd9523fa250bede0c30de8a4f1c983e58
tree9dd9a4a6a7ce0511ebab15c2908e0690af1652d0
parent0e36114e243888eecc16a693a17c4a0f7e9933f7
CUDA: fix FP16 overflow in tile FA kernel (llama/17875)
src/ggml-cuda/fattn-tile.cuh