mmarch/text-generation-webui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Avoid redundant function call in llamacpp_hf

Browse source
This commit is contained in:
oobabooga 2023-09-19 14:14:40 -07:00
parent 9b7646140c
commit 029da9563f
1 changed files with 3 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

5
modules/llamacpp_hf.py
View file

@ -131,9 +131,10 @@ class LlamacppHF(PreTrainedModel):
longest_prefix = min_length longest_prefix = min_length
if longest_prefix > 0: if longest_prefix > 0:
self.model.n_tokens = longest_prefix
self.model.eval(seq[longest_prefix:])
reset = False reset = False
self.model.n_tokens = longest_prefix
if len(seq_tensor) - longest_prefix > 0:
self.model.eval(seq[longest_prefix:])
if reset: if reset:
self.model.reset() self.model.reset()