Hey everyone
I am currently developing my own langchain application, taking inspiration from the newest short course ‘Langchain Chat with your own Data’.
I’ve been attempting to take this a step further by including a feature that allows the chatbot to access numerical databases and answer these data-related questions in a conversational way similar to the short course, by making plots, tables, manipulations, etc. Currently, I am experimenting with the pandas dataframe toolkit agent offered by Langchain. I’ve encountered several problems though as this agent toolkit lacks memory capabilities and isn’t compatible with the document model proposed in the short course.
If anyone has some tips on how to effectively implement this chatbot with the inclusion of numerical database access, ensuring it retains memory capabilities and compatibility with the document model, this would be greatly appreciated!
(a specific question: Can I still use the toolkit agent or should I make my own custom agent?)
Thank you in advance for your assistance!