Getting Error : LLM compiler architecture implementation using ChatOllama instead of ChatOpenAI

Getting error when using ChatOllama instead of ChatOpenAI for creating llm compiler agent using math tools. Does the current ChatOllama does not support few things or what is the problem?