Hi,
I am trying to model the Iris dataset using CrewAI. I have an agent to read the input file data, then pass the table to a data science agent who is to create and execute a classification neural network model to model the data, and then output the results. Is this possible without having to write the NN as a function for the agent to call? Instead I am interested to see if the agent can do this all by itself.
I have been trying, however it really struggles to write and execute the code by itself. It would be nice if there was a cookbook somewhere for how to leverage the allow_code_execution ability of the agents properly.
Thx