arXivBaohao Liao, Hanze Dong, Christof Monz, Xinxing Xu, Li Dong, Furu WeiMon, Jul 6, 2026, 12:56 AM PDT
score 16.7
New method speeds up AI training by reusing past agent interactions
Original: Multi-Turn On-Policy Distillation with Prefix Replay
Source: arxiv.org ↗
Writing ELI5 summary…