Hello

I'm trying to create an application using codellama. Basically a user inputs a dataset and then there are certain action they can make the llm perform. The problem I am encountering is that I am creating a dataframe and inputting the dataset in the llm prompt using df.to_string(). I think codellama isn't able to access the data properly since its answers are going haywire and none of them are related to the original dataset.

​

Has anyone worked with codellama who would be able to help me in this situation, or anybody who can help with alternative ways to input the dataset through which the LLM can effectively access it?

Any help would be appreciated.