apunkt/plano
1
0
Fork 0
mirror of https://github.com/katanemo/plano.git synced 2026-06-17 15:25:17 +02:00
Code Issues Projects Releases Packages Wiki Activity

Merge branch 'shuguang/main' of https://github.com/katanemo/arch into cotran/intent

Browse source
This commit is contained in:
cotran 2024-12-06 14:34:55 -08:00
commit f7d69d52a7
5 changed files with 37 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

28
model_server/app/model_handler/function_calling.py
View file

@ -99,20 +99,24 @@ class ArchIntentHandler(ArchBaseHandler):
Currently only support vllm inference
"""
messages = self._process_messages(
req.messages, req.tools, self.extra_instruction
)
# In the case that no tools are available, simply return `No` to avoid making a call
if len(req.tools) == 0:
model_response = Message(content="No", tool_calls=[])
else:
messages = self._process_messages(
req.messages, req.tools, self.extra_instruction
)
model_response = self.client.chat.completions.create(
messages=messages,
model=self.model_name,
stream=False,
extra_body=self.generation_params,
)
model_response = self.client.chat.completions.create(
messages=messages,
model=self.model_name,
stream=False,
extra_body=self.generation_params,
)
model_response = Message(
content=model_response.choices[0].message.content, tool_calls=[]
)
model_response = Message(
content=model_response.choices[0].message.content, tool_calls=[]
)
chat_completion_response = ChatCompletionResponse(
choices=[Choice(message=model_response)], model=self.model_name