tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-27 03:33:46 -04:00
Code Issues Packages Projects Releases Wiki Activity

kv-cache : allow context shift for recurrent models

Browse Source
This commit is contained in:
Francis Couture-Harpin
2025-05-02 19:29:23 -04:00
parent d55b0d0621
commit e94f3932f2
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
src/llama-kv-cache.cpp
View File

@ -1938,7 +1938,8 @@ llama_pos llama_kv_cache_recurrent::get_pos_max() const {
} }
bool llama_kv_cache_recurrent::get_can_shift() const { bool llama_kv_cache_recurrent::get_can_shift() const {
return false; // shifting is trivial, the recurrent states don't care about the absolute position
return true;
} }
uint32_t llama_kv_cache_recurrent::cell_max() const { uint32_t llama_kv_cache_recurrent::cell_max() const {