Smaller, weaker, yet better: Training llm reasoners via compute-optimal sampling

H Bansal, A Hosseini, R Agarwal, VQ Tran… - arxiv preprint arxiv …, 2024 - arxiv.org
Training on high-quality synthetic data from strong language models (LMs) is a common
strategy to improve the reasoning performance of LMs. In this work, we revisit whether this …

Rear: A relevance-aware retrieval-augmented framework for open-domain question answering

Y Wang, R Ren, J Li, WX Zhao, J Liu… - arxiv preprint arxiv …, 2024 - arxiv.org
Considering the limited internal parametric knowledge, retrieval-augmented generation
(RAG) has been widely used to extend the knowledge scope of large language models …