Post-Completion Learning for Language Models

Xiang Fei, Siqi Wang, Shu Wei +5

Current language model training paradigms typically terminate learning upon reaching the end-of-sequence (<eos>) token, overlooking the potential learning opportunities in the post…