Skip to content

Latest commit

 

History

History
44 lines (29 loc) · 1.28 KB

README.md

File metadata and controls

44 lines (29 loc) · 1.28 KB
Raw

Description

Intended to generate fictional test data for a school project using ChatGPT API, e.g.

Movie Name Movie Description Movie Genre
Catman Catman does Catman things Action

Name is a reference to Mockaroo

Quickstart

Create a .env file containing below

OPENAI_API_KEY=<KEY>

Make sure you have balance in your OpenAI account

Troubleshooting

1. Keep getting an error "You exceeded your current quota ..." despite having balance

If recently topped-up, recreate the API key

TODO

  1. Turn into a site:

    1. Password protect for cost reasons
    2. 3 Inputs:
      1. A model dropdown (at runtime, do a GET /models, list all gpt-*)
      2. Row number
      3. List of columns
    3. Show progress bar
    4. Downloads CSV when done
    5. Host the site in rpi, configure reverse proxy

  2. if large number of CSV, batch the calls to OpenAI then combine results at the end

    1. At the site, if batching, progress bar increases incrementally (e.g. 5 batches: 0% -> 20% -> 40% etc)
    2. 40 seems to be the max? maybe batch every 20 rows

  3. STRETCH GOAL: Allow uploading MySQL schemas and generate full e2e test data