# askai

**Repository Path**: alonegg/askai

## Basic Information

- **Project Name**: askai
- **Description**: jupyter magic
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-03-25
- **Last Updated**: 2023-03-25

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

ask_ai
================

<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->

A simple library for talking to ChatGPT from within Jupyter Notebook.

**Motivation**: *Don’t fear ChatGPT taking your job. Fear a person using
ChatGPT taking your job.*

Of course, this is a sensationalized statement. But it very succinctly
gets at something deep and important – [using LLMs in your work can
unlock massive productivity
gains](https://twitter.com/emollick/status/1631397931604488194?s=20). I
do not know of a better way for me to start using these tools than to
bring them closer to where I work, which is Jupyter Notebooks.

Also, I really hate the “typewriter effect” where you are fed a few
words at a time through the ChatGPT UI. I’d much rather have the full
reply I can scan in a couple of seconds.

## Install

1.  `pip install git+https://github.com/radekosmulski/ask_ai.git`
2.  Set the environment variable `OPENAI_API_KEY` to your OpenAI API
    key.
3.  Load the magics in your jupyter notebook: `%load_ext ask_ai.magics`

## How to use

Use `%%ai_ask` to start a new conversation, and `%%ai_continue` to
continue an existing conversation.

`%%ai_ask` starts a conversation with new (blank) context.

As long as you using `%%ai_continue` all (or the most recent) previous
messages will be forwarded as context. This can be helpful if your
question is related to what you have been discussing so far. In other
cases, it is faster and cheaper to use `%%ai_ask` as fewer tokens are
used.