This is regarding Agentic framework. We are getting issues while invoking the lang chain agent and need your help to get some suggestions from your expert team to resolve this.
As per our observation Agent is behaving bit weird while calling tools registered in UC :
Code Snippet :
prompt = get_prompt()
tools = get_tools()
agent = create_openai_tools_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
agent_str = ({ "messages": itemgetter("messages")} | agent_executor | itemgetter("output") | StrOutputParser())
- If the user prompt/demands can be fulfilled with invoking a single tool ( successful )
2. But if the user input demands chaining of tools to be invoked ,sometimes it’s throwing the following error. ( Error )[ But in some run the same code is working and returning valid output]
Cluster Used :
- ML Cluster runtime 15.4
- SQl Warehouse
