Observational Scaling Laws and the Predictability of Language Model Performance

Open in new window