StanGirard/quivr
1
1
Fork 0
mirror of https://github.com/StanGirard/quivr.git synced 2024-12-24 03:41:56 +03:00
Code Issues Packages Projects Releases Wiki Activity

fix(qa_base): asign max_token to llm (#1179)

Browse Source
This commit is contained in:
Gustavo J. Maciel 2023-09-15 12:51:53 -04:00 committed by GitHub
parent 980a704002
commit cdf587cfde
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 8 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
backend/llm/qa_base.py
View File

@ -94,7 +94,7 @@ class QABaseBrainPicking(BaseBrainPicking):
)
def _create_llm(
self, model, temperature=0, streaming=False, callbacks=None
self, model, temperature=0, streaming=False, callbacks=None, max_tokens=256
) -> BaseLLM:
"""
Determine the language model to be used.
@ -105,11 +105,12 @@ class QABaseBrainPicking(BaseBrainPicking):
"""
return ChatLiteLLM(
temperature=temperature,
max_tokens=max_tokens,
model=model,
streaming=streaming,
verbose=False,
callbacks=callbacks,
openai_api_key=self.openai_api_key,
openai_api_key=self.openai_api_key
) # pyright: ignore reportPrivateUsage=none
def _create_prompt_template(self):
@ -211,7 +212,7 @@ class QABaseBrainPicking(BaseBrainPicking):
self.callbacks = [callback]
answering_llm = self._create_llm(
model=self.model, streaming=True, callbacks=self.callbacks
model=self.model, streaming=True, callbacks=self.callbacks, max_tokens=self.max_tokens
)
# The Chain that generates the answer to the question

8
backend/routes/chat_routes.py
View File

@ -298,10 +298,10 @@ async def create_stream_question_handler(
else "gpt-3.5-turbo", # type: ignore
max_tokens=(brain_details or chat_question).max_tokens
if current_user.openai_api_key
else 0, # type: ignore
else 256, # type: ignore
temperature=(brain_details or chat_question).temperature
if current_user.openai_api_key
else 256, # type: ignore
else 0, # type: ignore
brain_id=str(brain_id),
user_openai_api_key=current_user.openai_api_key, # pyright: ignore reportPrivateUsage=none
streaming=True,
@ -314,10 +314,10 @@ async def create_stream_question_handler(
else "gpt-3.5-turbo",
temperature=chat_question.temperature
if current_user.openai_api_key
else 256,
else 0,
max_tokens=chat_question.max_tokens
if current_user.openai_api_key
else 0,
else 256,
user_openai_api_key=current_user.openai_api_key, # pyright: ignore reportPrivateUsage=none
chat_id=str(chat_id),
streaming=True,