baalajimaestro/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Remove usage of eval_tokens for cache check

Browse source
This commit is contained in:
Andrei Betlen 2023-05-26 20:12:05 -04:00
parent fe331ec589
commit 7fc7bc30e7
1 changed files with 2 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
llama_cpp/llama.py
View file

@ -735,10 +735,10 @@ class Llama:
try:
cache_item = self.cache[prompt_tokens]
cache_prefix_len = Llama.longest_token_prefix(
cache_item.eval_tokens, prompt_tokens
cache_item.input_ids.tolist(), prompt_tokens
)
eval_prefix_len = Llama.longest_token_prefix(
self.eval_tokens, prompt_tokens
self._input_ids.tolist(), prompt_tokens
)
if cache_prefix_len > eval_prefix_len:
self.load_state(cache_item)