Tag
TeachMateGPT presents a multi-agent framework that improves retrieval-augmented generation for creating pedagogical assessments from science textbooks, achieving higher faithfulness and answer relevancy compared to baseline systems.
This paper introduces Regret Pre-training, a self-supervised framework that uses a dual-view architecture to incorporate future context into causal language model training, improving performance on downstream tasks by up to 18 percentage points without adding parameters.