fix: Fix recurrent cache get after rebase

gabe-l-hart · gabe-l-hart · commit e053aad4c81a · 2025-06-19T09:23:05.000-06:00
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
@@ -13251,7 +13251,7 @@ 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
`@@ -13251,7 +13251,7 @@ struct llm_build_hybrid_mamba : public llm_graph_context {`
`13251`	`13251`	`}`
`13252`	`13252`
`13253`	`13253`	`// Extract the recurrent cache from the hybrid parent`
`13254`		`- const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid *>(memory)->get_child_cache<llama_kv_cache_recurrent>();`
	`13254`	`+ const auto * kv_recurrent = static_cast<const llama_kv_cache_hybrid_recurrent *>(memory)->get_kv_recurrent();`
`13255`	`13255`	`GGML_ASSERT(kv_recurrent);`
`13256`	`13256`
`13257`	`13257`	`for (int il = 0; il < n_layer; ++il) {`