Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
ggml: force flash attention off for grok
rickebbce136c7e0
diff --git a/fs/ggml/ggml.go b/fs/ggml/ggml.goindex 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}