Receive None from tool calling with Chatollama models

[GZCaz99]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_ollama import ChatOllama
from pydantic import BaseModel, Field



llm = ChatOllama(
        temperature=0.5,
        base_url="http://0.0.0.0:11434",
        model="llama3.1:latest",
    )


class Multiply(BaseModel):
    a: int = Field(..., description="First integer")
    b: int = Field(..., description="Second integer")

ans = llm.invoke("What is 45*67")
print(ans)
print(ans.tool_calls)

Description

So I've tried follow the tutorials from langchain , and I've also try creating my own tools, however. all the results from tool calls that I received is None. I've tried .invoke() method on a clean llm directly and it words fine, I've tried using different models deployed on the server but the tool_calls result is also None. In conclusion as long as i am using tool bonded llm, the response content will be empty and the tool_calls() result is None. I've looked up the langchain docs and find that ChatOllama is supported with tool calling , so as the model llama 3.1 im using. So I really wonder why.

System Info

Name: langchain
Version: 0.3.13
Name: langchain-ollama
Version: 0.2.2