5f8e2fa1718d1bbcadf1cd9c7a54fb8c-AuthorFeedback.pdf
–Neural Information Processing Systems
Comparison with other works, however,shows that scaling up isnecessary but not42 sufficient: our model gets better results compared to models with similar (or larger) size, batch-size, and compute43 requirements: Menicketal.
Neural Information Processing Systems
Feb-12-2026, 08:24:04 GMT