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

Fix llama.cpp double decoding

Browse source
This commit is contained in:
oobabooga 2023-09-17 13:07:48 -07:00
parent 230b562d53
commit d9b0f2c9c3
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
modules/llamacpp_model.py
View file

@ -116,7 +116,7 @@ class LlamaCppModel:
# Handle truncation # Handle truncation
prompt = self.encode(prompt) prompt = self.encode(prompt)
prompt = prompt[-get_max_prompt_length(state):] prompt = prompt[-get_max_prompt_length(state):]
prompt = self.decode(prompt).decode('utf-8') prompt = self.decode(prompt)
logit_processors = LogitsProcessorList() logit_processors = LogitsProcessorList()
if state['ban_eos_token']: if state['ban_eos_token']: