llama.cpp
Extend llama_kv_cache_seq_rm to allow matching any sequence
#3843
Merged

Loading