b1445
KerfuffleV2
tagged this
29 Oct 17:31
* Extend llama_kv_cache_seq_rm to allow matichng any sequence * Replace llama_kv_cache_tokens_rm with llama_kv_cache_clear Use llama_kv_cache_clear for cache clearing Change calls to llama_kv_cache_tokens_rm that want to delete by position to use llama_kv_cache_seq_rm functionality