]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
vulkan: Apply bias before softmax in FA, to avoid overflow (#24909)
authorJeff Bolz <redacted>
Wed, 24 Jun 2026 03:34:00 +0000 (22:34 -0500)
committerGitHub <redacted>
Wed, 24 Jun 2026 03:34:00 +0000 (22:34 -0500)
commitac4105d68b2955027115cf9bb50941ccf56974eb
tree1d35cfdd506bf57c61e6975cc67f0ac9372e9bb7
parentbe4a6a63eb2b848e19c277bdcf2bd399e8af76d9
vulkan: Apply bias before softmax in FA, to avoid overflow (#24909)
ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp
ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp