fix: Fix recurrent cache get after rebase

gabe-l-hart · gabe-l-hart · commit 585444b7bf44 · 2025-05-28T13:28:09.000-06:00
Branch: GraniteFour

Signed-off-by: Gabe Goodhart &lt;ghart@us.ibm.com&gt;
diff --git a/src/llama-model.cpp b/src/llama-model.cpp
@@ -12977,7 +12977,7 @@ struct llm_build_hybrid_mamba : public llm_graph_context {
         }
 
         // Extract the recurrent cache from the hybrid parent
-        const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid *>(memory)->get_child_cache<llama_kv_cache_recurrent>();
+        const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid_recurrent *>(memory)->get_kv_recurrent();
         GGML_ASSERT(kv_recurrent);
 
         for (int il = 0; il < n_layer; ++il) {

Original file line number	Diff line number	Diff line change
`@@ -12977,7 +12977,7 @@ struct llm_build_hybrid_mamba : public llm_graph_context {`
`12977`	`12977`	`}`
`12978`	`12978`
`12979`	`12979`	`// Extract the recurrent cache from the hybrid parent`
`12980`		`- const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid *>(memory)->get_child_cache<llama_kv_cache_recurrent>();`
	`12980`	`+ const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid_recurrent *>(memory)->get_kv_recurrent();`
`12981`	`12981`	`GGML_ASSERT(kv_recurrent);`
`12982`	`12982`
`12983`	`12983`	`for (int il = 0; il < n_layer; ++il) {`