Taming the Long-Tail: Efficient Reasoning RL Training with Adaptive Drafter, [ASPLOS 2026]
Paper: https://arxiv.org/abs/2511.16665 Github: https://github.com/mit-han-lab/fastrl
Paper: https://arxiv.org/abs/2511.16665 Github: https://github.com/mit-han-lab/fastrl