Author name: Amit Kiran Rege

arXiv:2604.04855v1 Announce Type: new
Abstract: We study how generator access constrains autoregressive post-training. The central question is whether the learner is confined to fresh root-start rollouts or can return to previously built prefixes and …

arXiv:2604.04892v1 Announce Type: new
Abstract: Machine learning models increasingly generate their own training data — online bandits, reinforcement learning, and post-training pipelines for language models are leading examples. In these adaptive se…