If you're doing RL post-training, rollout generation dominates wall-clock time; distribution-aware speculative decoding adapts the draft/verify process to the evolving policy distribution to cut rollout time without hurting reward quality, unlike static speculative decoding tuned for fixed inference workloads.
DAS fixes it with adaptive speculative decoding — up to 50% faster, zero degradation in reward quality.
Transcript
Rollout is the silent bottleneck in RL post-training. DAS fixes it with adaptive speculative decoding — up to 50% faster, zero degradation in reward quality.
