method · first seen 7 Nov 2026
reinforcement learning rollouts
Generating sampled trajectories from a language model during on-policy RL training, requiring reproducible inference.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Detent: Batch-Invariant LLM Serving and RL Rollouts at Near-Stock SpeedPapers · 7 Nov 2026 · Peixu Xin, Runchuan Feng, Jiazhen Guo, Kan Wu, Ming Yin
Latest sources
- Detent: Batch-Invariant LLM Serving and RL Rollouts at Near-Stock SpeedPapers · 7 Nov 2026 · Peixu Xin, Runchuan Feng, Jiazhen Guo, Kan Wu, Ming Yin