SRT: Accelerating Reinforcement Learning via Speculative Rollout with Tree-Structured Cache

Documenti analoghi

• Speculate Deep and Accurate: Lossless and Training-Free Acceleration for Offloaded LLMs via Substitute Speculative Decoding
  di: Wang, Pei-Shuo, et al.
  Pubblicazione: (2025)
• DisagMoE: Computation-Communication overlapped MoE Training via Disaggregated AF-Pipe Parallelism
  di: Zeng, Zhichen, et al.
  Pubblicazione: (2026)
• SPEC-RL: Accelerating On-Policy Reinforcement Learning with Speculative Rollouts
  di: Liu, Bingshuai, et al.
  Pubblicazione: (2025)
• OpenTinker: Separating Concerns in Agentic Reinforcement Learning
  di: Zhu, Siqi, et al.
  Pubblicazione: (2026)
• Accelerating Large-Scale Reasoning Model Inference with Sparse Self-Speculative Decoding
  di: Zhao, Yilong, et al.
  Pubblicazione: (2025)