Skip to content

armgit / camel

Public Imported from GitHub
0

🐫 CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org

bug fix: vllm can't do tool call (#1829)

Rishabha4f487012961
diff --git a/camel/models/vllm_model.py b/camel/models/vllm_model.py
index 353655e6f..200c5d2fe 100644
--- a/camel/models/vllm_model.py
+++ b/camel/models/vllm_model.py
@@ -139,6 +139,10 @@ class VLLMModel(BaseModelBackend):
Args:
messages (List[OpenAIMessage]): Message list with the chat history
in OpenAI API format.
+ response_format (Optional[Type[BaseModel]], optional): The format
+ to return the response in.
+ tools (Optional[List[Dict[str, Any]]], optional): List of tools
+ the model may call.
Returns:
Union[ChatCompletion, AsyncStream[ChatCompletionChunk]]:
@@ -146,10 +150,16 @@ class VLLMModel(BaseModelBackend):
`AsyncStream[ChatCompletionChunk]` in the stream mode.
"""
+ kwargs = self.model_config_dict.copy()
+ if tools:
+ kwargs["tools"] = tools
+ if response_format:
+ kwargs["response_format"] = {"type": "json_object"}
+
response = await self._async_client.chat.completions.create(
messages=messages,
model=self.model_type,
- **self.model_config_dict,
+ **kwargs,
)
return response
@@ -164,6 +174,10 @@ class VLLMModel(BaseModelBackend):
Args:
messages (List[OpenAIMessage]): Message list with the chat history
in OpenAI API format.
+ response_format (Optional[Type[BaseModel]], optional): The format
+ to return the response in.
+ tools (Optional[List[Dict[str, Any]]], optional): List of tools
+ the model may call.
Returns:
Union[ChatCompletion, Stream[ChatCompletionChunk]]:
@@ -171,10 +185,16 @@ class VLLMModel(BaseModelBackend):
`Stream[ChatCompletionChunk]` in the stream mode.
"""
+ kwargs = self.model_config_dict.copy()
+ if tools:
+ kwargs["tools"] = tools
+ if response_format:
+ kwargs["response_format"] = {"type": "json_object"}
+
response = self._client.chat.completions.create(
messages=messages,
model=self.model_type,
- **self.model_config_dict,
+ **kwargs,
)
return response