Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding
This work provides a practical system integration of speculative decoding to accelerate RL post-training for large language models, a critical bottleneck for frontier model development.
The paper addresses the rollout generation bottleneck in RL post-training of language models by integrating speculative decoding as a lossless acceleration primitive. They achieve 1.8x rollout throughput improvement at 8B scale in synchronous RL and project up to 2.5x end-to-end training speedup at 235B scale with asynchronous RL.
RL post-training of frontier language models is increasingly bottlenecked by autoregressive rollout generation, making rollout acceleration a central systems challenge. Many existing efficiency methods improve throughput by changing the rollout or optimization regime, for example, through off-policy execution, replay, or lower-precision generation. We study speculative decoding as a lossless acceleration primitive for RL rollouts that preserves the target model's output distribution. We implement speculative decoding in NeMo-RL with a vLLM backend, supporting both synchronous and asynchronous pipelines and enabling speculation during RL rollouts. This benefit is realizable across speculation mechanisms, such as pretrained MTP heads, small external draft models or even techniques such as Eagle3, which are traditionally applied after RL phase. This yields a deployment path for state-of-the-art speculative decoding inside RL training. In a reasoning post-training workload at 8B scale under synchronous RL, speculative decoding improves rollout throughput by 1.8x. Using a high-fidelity performance simulator, we project that combining speculative decoding with asynchronous RL yields up to 2.5x end-to-end training speedup at 235B scale.