Lacuna
Author of "Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning" on arXiv
How media typically covers Weiran He
Weiran He as author
Seer, an online context learning system, improves synchronous LLM reinforcement learning throughput by 74-97% and reduces long-tail latency by 75-93% through divided rollout, context-aware scheduling, and adaptive speculative decoding.
“Author of "Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning" on arXiv”