arXivYongliang Miao, Fengyuan Liu, Wei Shi, Yanguang Liu, Fei Sun, Na Zou, Mengnan DuFri, Jun 5, 2026, 12:52 AM PDT
score 15.2
Adaptive training lets AI learn from examples without copying blindly
Original: RASFT: Rollout-Adaptive Supervised Fine-Tuning for Reasoning
Source: arxiv.org ↗
Writing ELI5 summary…