English(EN)Quantifying Empirical Compute-Supervision Tradeoffs in RLVR
新的RLVR方法提高了LLM的训练效率和数据选择
作者PulseAugur 编辑部·[8 个来源]·
研究人员正在开发新的方法来提高用于训练大型语言模型（LLM）的可验证奖励强化学习（RLVR）的效率和有效性。两篇论文介绍了新颖的数据选择技术：SHIFT，它使用推理时的隐藏状态动态来选择实例而无需事先训练；IRDS，它采用与验证器耦合的稀疏自动编码器来进行可审计的实例选择。另一项研究调查了RLVR中计算与监督质量之间的权衡，发现验证器质量，特别是减少假阴性，比单独扩展计算更关键。最后，提出了一种时间调度方法来优化随时间的学习信号，从而实现更稳定和高效的策略演进。
AI
arXiv:2605.28631v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) can yield large reasoning gains from very few training instances, yet its strong sensitivity to which instances are used makes data selection a central bottleneck. Most existing …
arXiv:2605.28247v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has become a key technique for en- hancing LLM reasoning, yet its data ineffi- ciency remains a major bottleneck. Existing methods address this problem only partially, each mis…
Reinforcement learning with verifiable rewards (RLVR) can yield large reasoning gains from very few training instances, yet its strong sensitivity to which instances are used makes data selection a central bottleneck. Most existing selection pipelines rely on training-time optimi…
arXiv:2605.26971v1 Announce Type: new Abstract: The proliferation of Reinforcement Learning from Verifiable Rewards (RLVR) datasets has exacerbated provenance collapse due to unclear lineage among existing datasets. To bridge this fragmented RLVR data landscape, we propose Atomic…
The proliferation of Reinforcement Learning from Verifiable Rewards (RLVR) datasets has exacerbated provenance collapse due to unclear lineage among existing datasets. To bridge this fragmented RLVR data landscape, we propose Atomic-source Tracing via Lineage-Aware Search (ATLAS)…
The proliferation of Reinforcement Learning from Verifiable Rewards (RLVR) datasets has exacerbated provenance collapse due to unclear lineage among existing datasets. To bridge this fragmented RLVR data landscape, we propose Atomic-source Tracing via Lineage-Aware Search (ATLAS)…
arXiv cs.AI
TIER_1English(EN)·Ryo Mitsuhashi, Patrick Chen, Isabelle Tseng, Jasin Cekinmez, Addison J. Wu·
arXiv:2605.25252v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has become a standard paradigm for post-training language models, but in practice, verifiers are rarely perfect. Recent theoretical work predicts that verifier noise affects th…
arXiv:2605.25381v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has become a core technique for post-training of Large Language Models (LLMs). While policy optimization is driven by all sampled tokens under a globally broadcast scalar reward,…
