用强化学习动态优化预训练数据配比，Actor-Critic框架让大模型学得更高效。
arXiv:2505.23878v2 Announce Type: replace-cross Abstract: Optimizing pretraining data composition is pivotal for LLM generalization. While dynamic mix…
ICML 2026 论文提出新型不对称演员-评论家框架，利用特权信号突破全局状态假设，大幅提升强化学习样本效率。
arXiv:2509.26000v3 Announce Type: replace Abstract: Asymmetric reinforcement learning leverages privileged information available during training to im…
用LLM智能体推理改进符号回归的样本效率，新方法区分候选生成与搜索引导。
arXiv:2606.04360v1 Announce Type: new Abstract: Symbolic regression (SR) discovers compact mathematical expressions from data, yet recent LLM-based ev…
被ICML2026接收的新方法，用评论家指导扩散模型实现样本高效的强化学习。
arXiv:2605.30056v1 Announce Type: cross Abstract: Recent advances in reinforcement learning (RL) have achieved great successes by leveraging the multi…
提出双层优化方法，让可学习的文本反馈高效引导LLM推理，解决稀疏奖励下的样本效率难题。
arXiv:2605.24547v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards can improve LLM reasoning, but learning remains sample-…
提出在任务无关世界模型中强化VLA的方法，提升模型泛化能力与样本效率。
arXiv:2605.12334v2 Announce Type: replace Abstract: Post-training Vision-Language-Action (VLA) models via reinforcement learning (RL) in learned world…
用随机少样本指导提升RLVR在困难问题上的样本效率，大模型训练新思路。
arXiv:2605.15012v1 Announce Type: cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has achieved great success in developing Large…