Inference-Aware Fine-Tuning for Best-of-NSampling in Large Language Models

https://arxiv.org/pdf/2412.15287