bump llama.cpp

sinkingsugar · sinkingsugar · commit e372e7be072f · 2025-03-14T08:57:36.000+08:00
diff --git a/deps/llama.cpp b/deps/llama.cpp
@@ -1 +1 @@
-Subproject commit 13763a2a9b967e1887f218909a670cd8cbc31842
+Subproject commit 830965b4fef3834f2809a354df240ec88fc2edf0
diff --git a/shards/modules/llm/embedd.cpp b/shards/modules/llm/embedd.cpp
@@ -334,7 +334,7 @@ struct Embed {
     }
     batch.n_tokens = tokens.size();
 
-    llama_kv_cache_clear(llmContext.ctx.get());
+    llama_kv_self_clear(llmContext.ctx.get());
 
     if (llama_model_has_encoder(model)) {
       if (llama_encode(llmContext.ctx.get(), batch) < 0) {

Original file line number	Diff line number	Diff line change
`@@ -334,7 +334,7 @@ struct Embed {`
`334`	`334`	`}`
`335`	`335`	`batch.n_tokens = tokens.size();`
`336`	`336`
`337`		`- llama_kv_cache_clear(llmContext.ctx.get());`
	`337`	`+ llama_kv_self_clear(llmContext.ctx.get());`
`338`	`338`
`339`	`339`	`if (llama_model_has_encoder(model)) {`
`340`	`340`	`if (llama_encode(llmContext.ctx.get(), batch) < 0) {`