After using Cursor to ask LLMs questions and brainstorm ideas, I wanted a similar experience in IPython to limit context switching between terminal and browser.
The extension finds the relevant context—column names, shape, and rows of the dataframe—and sends that alongside the IPython history to the LLM for more relevant responses.
I’m also planning to add support for local models to compare them with Claude 3.5 Sonnet, which I believe generates the best responses for coding questions.
You can install ipychat using pip and configure it based on which LLM model you want to use, enhancing the data exploration experience.
Collection
[
|
...
]