Skip to yearly menu bar Skip to main content


For Perception Tasks: The Cost of LLM Pretraining by Next-Token Prediction Outweigh its Benefits

Randall Balestriero · Hai Huang

Abstract

Chat is not available.