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

Bugfix: only eval new tokens

Browse source
This commit is contained in:
Andrei Betlen 2023-04-15 17:32:53 -04:00
parent 887f3b73ac
commit 89856ef00d
1 changed files with 1 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
llama_cpp/llama.py
View file

@ -280,6 +280,7 @@ class Llama:
if self.verbose: if self.verbose:
print("generate cache hit", file=sys.stderr) print("generate cache hit", file=sys.stderr)
reset = False reset = False
tokens = tokens[len(self.tokens) :]
### ###
if reset: if reset:
self.reset() self.reset()