Find answers from the community

Updated 2 years ago

moreover when I used the code for the

At a glance

moreover when I used the code for the chatbot I noticed that it used text-davinci model even if I indicate the agent to use gpt3.5.

7 comments

jjerryjliu0

yeah you will want to specify an llm predictor

jjerryjliu0

so for instance when specifying an IndexToolConfig:

Plain Text

tool_config = IndexToolConfig(
        index=index_set[y], 
        name=f"Vector Index {y}",
        description=f"useful for when you want to answer queries about the {y} SEC 10-K for Uber",
        index_query_kwargs={"similarity_top_k": 3, "llm_predictor": llm_predictor},
        tool_kwargs={"return_direct": True}
    )

jjerryjliu0

notice the "llm_predictor" in index_query_kwargs

pplouplou

Oh yeah I'm so stupid x) Also, thank you so much for your time and all the work you are doing with your team that's really helpful, amazing !

pplouplou

I'm using this line --> index_query_kwargs={"similarity_top_k": 1, "llm_predictor": llm} with llm=OpenAIChat(temperature=0.0, model_name="gpt-3.5-turbo") but now got this error : AttributeError: 'ChatOpenAI' object has no attribute 'total_tokens_used'

pplouplou

I don't get it

jjerryjliu0

can you paste your code? the one thing is you should use ChatOpenAI and not OpenAIChat

Add a reply