扩散与流模型的黑盒对齐新方法，信任区域噪声搜索带来高效微调
arXiv:2603.14504v2 Announce Type: replace-cross Abstract: Optimizing the noise samples of diffusion and flow models is an increasingly popular approac…
将信任区域优化引入策略蒸馏，解决模型压缩中的策略漂移问题。
arXiv:2607.04751v1 Announce Type: cross Abstract: Big goals are hard to achieve all at once; breaking them into small steps is wiser. We present Trust…
用贝叶斯锚定潜在信任区域优化高维不确定下的结构设计分类问题，突破传统贝叶斯优化局限
arXiv:2604.25241v2 Announce Type: replace Abstract: Categorical structural optimization under aleatoric uncertainty is challenging because each design…
新方法Trust Region Masking，通过动态约束策略更新提升大模型长序列强化学习稳定性与收敛性。
arXiv:2512.23075v5 Announce Type: replace-cross Abstract: Policy gradient methods for Large Language Models optimize a policy $\pi_\theta$ via a surro…
非均匀令牌级信任区域优化，突破传统限制提升大模型强化学习训练稳定性。
arXiv:2606.10968v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has become standard for improving LLM reasonin…
融合信任区域与在线策略的蒸馏方法，为强化学习知识迁移提供新思路。
arXiv:2606.01249v1 Announce Type: new Abstract: On-Policy Distillation (OPD) is a fundamental technique for efficient post-training of large language …
在线策略蒸馏新突破，信任区域行为混合让策略学习更稳定高效。
arXiv:2605.31159v1 Announce Type: cross Abstract: On-policy distillation (OPD) trains a student on prefixes sampled from its own policy while matching…
将信任区域优化与元学习思想巧妙融合，提出持续学习新范式，在防止灾难性遗忘的同时实现高效知识迁移，实验数据丰富。
arXiv:2602.02417v2 Announce Type: replace Abstract: Continual learning aims to acquire tasks sequentially without catastrophic forgetting, yet standar…
强化学习新突破：信任区域与Q学习的巧妙融合，伴随匹配算法提升训练稳定性
arXiv:2605.27079v1 Announce Type: cross Abstract: Off-policy reinforcement learning of pretrained flow policies remains challenging due to the instabi…
重新审视LLM强化学习中的信任区域，提出新颖的优化思路，值得关注。
arXiv:2602.04879v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become a cornerstone for fine-tuning Large Language Models (…
提出先验校正正交信任区域引导，优化动作分块流策略，平滑性与性能兼得。
arXiv:2605.24433v1 Announce Type: cross Abstract: Flow-matching robot policies commonly use action-chunking inference for efficient closed-loop contro…
提出Delta Forcing方法，解决交互式自回归视频生成中响应性与稳定性的平衡难题。
arXiv:2605.14382v2 Announce Type: replace Abstract: Interactive real-time autoregressive video generation is essential for applications such as conten…
针对多智能体LLM协调中序列微调导致上下文分布偏移的缺陷，提出信任区域微调方法，有效提升团队协同表现。
arXiv:2605.15207v1 Announce Type: new Abstract: Multi-agent LLM systems have shown promise for complex reasoning, yet recent evaluations reveal they o…