Awesome, anyone have specific ideas what this can solve? I'm at a loss for specifics, I don't deal much with biology.

This is more awesome than the other news here lol

If you read through the lines they "just" used GPT 2.0 (!!!) to beat all other models with their first run in September 22.

Old models were trained on medical literature (and case studies) in order to produce conclusions for specific sub-medical fields (oncology, neurology, etc.). BioGPT is one of the first generalized models that can produce results for all fields without constraints and beat the old models in their pre-trained domain.

When can I use it?

This is my question too.

I am stupid for not learning more about coding but can this be locally run? Seems like a big model from what I can tell. Is the github code the whole model?

Maybe I just need to dust off the old learning chops to answer all those questions myself but I'd alsp love to give this a try in some form as soon as possible.