Merge branch 'xsn/gemma3_text' into xsn/gemma3_vision

ngxson · ngxson · commit 5cf1cc5210d8 · 2025-03-12T08:06:34.000+01:00
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
@@ -879,8 +879,8 @@ void llama_model::load_hparams(llama_model_loader & ml) {
                 }
 
                 hparams.f_attention_scale = type == LLM_TYPE_27B
-                    ? 1.0f / std::sqrtf(float(hparams.n_embd / hparams.n_head(0)))
-                    : 1.0f / std::sqrtf(float(hparams.n_embd_head_k));
+                    ? 1.0f / std::sqrt(float(hparams.n_embd / hparams.n_head(0)))
+                    : 1.0f / std::sqrt(float(hparams.n_embd_head_k));
             } break;
         case LLM_ARCH_STARCODER2:
             {

Original file line number	Diff line number	Diff line change
`@@ -879,8 +879,8 @@ void llama_model::load_hparams(llama_model_loader & ml) {`
`879`	`879`	`}`
`880`	`880`
`881`	`881`	`hparams.f_attention_scale = type == LLM_TYPE_27B`
`882`		`- ? 1.0f / std::sqrtf(float(hparams.n_embd / hparams.n_head(0)))`
`883`		`- : 1.0f / std::sqrtf(float(hparams.n_embd_head_k));`
	`882`	`+ ? 1.0f / std::sqrt(float(hparams.n_embd / hparams.n_head(0)))`
	`883`	`+ : 1.0f / std::sqrt(float(hparams.n_embd_head_k));`
`884`	`884`	`} break;`
`885`	`885`	`case LLM_ARCH_STARCODER2:`
`886`	`886`	`{`