Hacker News new | ask | show | jobs
by minimaxir 2673 days ago
The code is out (https://github.com/openai/gpt-2), just not the larger model itself.
1 comments

Saying 'the code is out' is a little misleading. It's some supporting code for working with a trained model, and a model definition of what is already a standard and widely-implemented architecture. It doesn't include the dataset, the code used to construct the dataset, or perhaps most importantly of all, the code to train a model on a fleet of TPUv3s.