AI训练提速3.8倍，但数据过期怎么办？
大模型训练有个瓶颈：生成训练数据（rollout）和更新模型必须同步，像流水线上一环卡一环。这篇论文发现，把两者拆开异步跑，速度能提升1.6到3.8倍，但代价是模型用的数据可能“过期”——就像用昨天的新闻训练今天的AI。研究者系统分析了这种“过期”问题，发现不同训练方法影响不同：一种叫前向KL的方法更抗过期，而反向KL则脆弱。他们尝试了各种修复方案，最终提出一个简单有效的办法：在模型更新时重新计算反向KL信号。此外，他们还解决了存储教师模型输出时精度与效率的权衡。这套异步训练框架已开源。它不是你明天就能直接用的工具，但揭示了AI训练效率与数据新鲜度之间的核心矛盾，对关注大模型训练成本的人有启发。
📄 原文摘要(英文)
On-policy distillation (OPD) trains a student on its own rollouts guided by teacher feedback and is becoming increasingly important for large language model (LLM) post-training. Like reinforcement learning (RL), however, OPD faces an on-policy systems bottleneck, as rollouts can dominate training time for reasoning workloads. Asynchronous training pipelines can alleviate this bottleneck by decoupling rollout generation from learner updates, but doing so introduces stale-policy data. While prior work has studied stale data in asynchronous RL, its effects in OPD remain underexplored. We present the first systematic study of staleness in asynchronous OPD, focusing on a practical setting where teacher feedback is implemented through local KL losses and full-vocabulary teacher logits are too expensive to store or transfer, necessitating finite teacher-score caches. We first show that KL direction changes the stale-data problem: teacher-weighted forward KL is more robust to stale rollouts, whereas student-weighted reverse KL is vulnerable. Second, for this vulnerable reverse-KL case, we study whether methods designed to stabilize asynchronous RL can mitigate OPD staleness. In our experiments, they do not improve over a simpler OPD-specific surrogate: recomputing the reverse-KL signal under the current student at learner time. Third, we analyze how finite teacher-score caches create a bias-variance tradeoff for sparse and sampled reverse-KL OPD estimators. This motivates multi-sample Monte Carlo (MC), which preserves MC correctability while reducing one-sample variance. Finally, we present and open-source AsyncOPD, a fully asynchronous OPD training pipeline built from these estimator choices. Experiments show that AsyncOPD improves training throughput by 1.6times to 3.8times over strict synchronous training while reaching comparable accuracy.