bobmaster/matrix_chatgpt_bot
1
0
Fork 0
Code Issues 5 Pull requests 2 Projects Releases 10 Packages Wiki Activity

Fallback to gpt-3.5-turbo when caculate tokens using custom model

Browse source
This commit is contained in:
hibobmaster 2023-12-23 21:16:37 +08:00
parent fac14a4244
commit 96a83fd824
Signed by: bobmaster
SSH key fingerprint: SHA256:5ZYgd8fg+PcNZNy4SzcSKu5JtqZyBF8kUhY7/k2viDk
1 changed files with 4 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
src/gptbot.py
View file

@ -122,13 +122,13 @@ class Chatbot:
"""
Get token count
"""
_engine = self.engine
if self.engine not in ENGINES:
raise NotImplementedError(
f"Engine {self.engine} is not supported. Select from {ENGINES}",
)
# use gpt-3.5-turbo to caculate token
_engine = "gpt-3.5-turbo"
tiktoken.model.MODEL_TO_ENCODING["gpt-4"] = "cl100k_base"
encoding = tiktoken.encoding_for_model(self.engine)
encoding = tiktoken.encoding_for_model(_engine)
num_tokens = 0
for message in self.conversation[convo_id]: