4 ms·
GPT-4 Trained on 10x More Data, Can Write 60k Word Books
- minimaxir 4y agoThis misinformed GPT-4 take is new levels of comically bad because it correlates number of model hyperparameters with the quantity of training data.
loading thread…
4 ms·