tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-05 00:25:26 -04:00
Code Issues Packages Projects Releases Wiki Activity

fix: Fix shift logic to defer to unified cache

Browse Source 
Branch: HybridRecurrentCache

Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
This commit is contained in:
Gabe Goodhart
2025-06-03 16:29:40 -06:00
parent 6c6ec0003a
commit cf03d4ae5c
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/llama-kv-cache-hybrid-recurrent.cpp
View File

@@ -150,8 +150,8 @@ void llama_kv_cache_hybrid_recurrent::defrag_sched(float thold) {
}
bool llama_kv_cache_hybrid_recurrent::get_can_shift() const {
// TODO: Should this return true if the attention cache can shift?
return false;
// Shifting is trivially supported for recurrent
return kv_attn->get_can_shift();
}
void llama_kv_cache_hybrid_recurrent::state_write(llama_io_write_i & io, llama_seq_id seq_id) const {