HACKER Q&A
📣 mupuff1234

Why are we not seeing “Large X models”?


Where X could be math/physics/biology/etc?


  👤 PaulHoule Accepted Answer ✓
They exist. People have trained BERT-like and other transformer models for specialist domains, particularly medicine. I “see” them all the time on arXiv.org.

Models like that perform well if you are doing classification, information extraction, etc.

For text generation though they have the same bullshitting problem as ChatGPT however, remember that science chatbot that Facebook released that they had to take down?