Large Language Model playground

[kevinodotnet]

Ottwatch collects a lot of data (and could collect a lot more), so while we have it all lying around, let's get funky:

• choose an open-source large-language model system
• training it, unsupervised, on the corpus of data available in ottwatch
  • dev apps, and their associated PDFs (to text)
  • lobbying data
  • city agendas and attached reports, etc
  • ...
• then create a chatbot that people can use to query the LLM

This might be way too costly to do in $ terms, but it could be fun to try it out