Skip to content

Latest commit

 

History

History
10 lines (7 loc) · 401 Bytes

README.md

File metadata and controls

10 lines (7 loc) · 401 Bytes

mess-gpt-preprocessor

FB Messenger transcription preprocessor to create data for fine-tuning ChatGPT.

Usage

  1. Run the scroll.js and download.js scripts to get the transcript.txt input file. (these are private for now, so don't look for them on my GH)
  2. Change the config in ./__main__.py.
  3. Run script: py ./__main__.py
  4. Upload output.jsonl to the OpenAI Fine-tuning API.