From ebbce136c7e035872ca7027dc9438bc2efc17123 Mon Sep 17 00:00:00 2001 From: rick Date: Wed, 25 Mar 2026 02:51:46 +0100 Subject: [PATCH] ggml: force flash attention off for grok --- fs/ggml/ggml.go | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/ggml/ggml.go b/fs/ggml/ggml.go index 53d1543c2..c835cb32b 100644 --- a/fs/ggml/ggml.go +++ b/fs/ggml/ggml.go @@ -874,7 +874,7 @@ func (f GGML) SupportsFlashAttention() bool { return true } - if slices.Contains([]string{"gemma2"}, arch) { + if slices.Contains([]string{"gemma2", "grok"}, arch) { return false }