My Obsidian Blog

HF Daily Papers 2026-07-28

16分钟阅读

HF Daily Papers 2026-07-28

范围：优先 RLHF / preference optimization / alignment 在 Diffusion 或生成式 Diffusion 中的应用；其次为 RLHF 在 VLM / LLM 中的工作。排除 VLA、机器人、具身智能、robot manipulation、coding / software-engineering agents、long-horizon agents 与 agentic-RL。

筛选说明

  • 本列表只基于 Hugging Face Daily Papers 可见元数据与摘要完成初筛和排序；不应将作者摘要中的主张视为已被独立验证的事实。
  • 中文 AI Summary 是摘要级判断，不等同于全文结论。
  • 相关论文数：5。
  • 全文精读候选数：2。

排序后的相关论文

1. ⭐ Rethinking Classifier-Free Guidance in On-Policy Diffusion Distillation

中文 AI Summary

本文探讨了在 Classifier-Free Guidance (CFG) 下对扩散模型进行在线策略蒸馏（On-Policy Distillation）时发生的负分支不对称（NBA）问题。作者提出了分支感知的 PDM 优化目标，分别约束正面预测与 CFG 条件方向。基于摘要判断，该方法显著提升了密集到稀疏视频控制蒸馏的鲁棒性。

Abstract

On-policy distillation (OPD) adapts diffusion models by querying a teacher along trajectories generated by the current student, but how it should behave under classifier-free guidance (CFG), a default component of modern diffusion systems, remains poorly understood. Existing OPD methods naturally extend velocity matching to the CFG-composed prediction, directly matching teacher and student guided velocities. We show that this objective is under-identified at the branch level: positive- and negative-branch errors can compensate in the guided prediction. Through two contrasting cases, we find that naive matching remains effective under shared negative conditioning, where both branch errors decrease jointly. When the model’s native CFG schema retains privileged information in the teacher’s negative branch that is unavailable to the student, however, this joint reduction breaks down and the composed objective induces antagonistic branch-error dynamics, reducing the positive-branch error while increasing the negative-branch error. We term this failure mode Negative Branch Asymmetry (NBA). To address NBA, we introduce Positive—Direction Matching (PDM), a branch-aware OPD objective that separately constrains the positive prediction and the CFG conditional direction. We apply PDM to dense-to-sparse video control, where naive guided matching is highly sensitive to inference guidance scales, while branch-aware supervision enables more robust and effective knowledge transfer.

排序依据与全文待核验点

  • 为什么相关： Directly addresses trajectory/branch-level error dynamics under classifier-free guidance (CFG) in on-policy diffusion distillation, proposing Positive-Direction Matching (PDM) to fix negative branch asymmetry.
  • 全文待核验： Verify the exact mathematical formulation of the Positive-Direction Matching (PDM) loss and its stability across different CFG scale hyper-parameters.

2. ⭐ Oxygen-TryOn: Fashion-Native Foundation Model for Any-item Virtual Try-On

  • 方向： RLHF-Diffusion
  • 研究阅读价值： 4/5
  • 潜在工程价值： 5/5
  • 作者： Yong Liu, Xiaolong Fu, Zihang Xu, Wen Xue, Xueheng Li, Lin Song, Yuan Zhang, Chuyang Zhao, Haoyang Huang, Nan Duan, Yipeng Sun, Yan Li, Simiu Gu
  • Hugging Face： 论文页
  • arXiv： 摘要页 · PDF
  • 代码： 未提供
  • 项目页： https://oxygenvision.github.io/Oxygen-TryOn/

中文 AI Summary

本文提出了 Oxygen-TryOn 虚拟试衣基础模型，采用了 CPT、SFT 和 RL 三阶段训练 Recipe。其 RL 阶段结合专用试衣 Reward Model 与通用评估模型进行强化学习微调。基于摘要判断，该方法在保持服装细节与人体姿态控制上达到了 SOTA。

Abstract

We present Oxygen-TryOn, a unified foundation model for any-item virtual try-on. Rather than repurposing a general-purpose image editor, Oxygen-TryOn is fashion-native, built for try-on through a dedicated data engine and try-on-specific training. Given one or more reference items (clean product shots or in-the-wild worn-on photos) and a single target subject image, it synthesizes a photorealistic image of the subject wearing the items across virtually any fashion category. Prior systems handle a single garment category in a studio setting, and recent multi-reference methods remain garment-centric; in contrast, Oxygen-TryOn supports diverse items and scenarios, including full- and half-body views, a variable number of references, and free multi-item composition, while faithfully preserving both subject identity and item appearance. Instead of mask-based inpainting, we reformulate try-on as a multi-reference, understanding-driven generation task. We build a data engine that collects, manufactures, annotates, and filters high-quality try-on data at scale, and design a three-stage recipe of continued pre-training (CPT), supervised fine-tuning (SFT), and reinforcement learning (RL). The RL stage uses a hybrid reward combining an in-house try-on reward model with a proprietary, rubric-guided general-purpose model, jointly supervising fine-grained consistency and instruction-level quality. It also follows general editing instructions (e.g., pose changes) in the same pass. Across public benchmarks and our in-house Oxygen-TryOn Bench, it achieves state-of-the-art consistency and realism on single-item try-on and leads on multi-item try-on, matching or surpassing both leading proprietary systems (Nano Banana Pro, GPT-Image-2, Seedream5 Lite) and open-source models (FLUX.2).

排序依据与全文待核验点

  • 为什么相关： Applies RL with a hybrid reward model (combining an in-house try-on reward model with a general rubric-guided model) directly to fine-tune a fashion/try-on image generation diffusion model.
  • 全文待核验： Verify the RL training formulation, loss weights, and reward model architecture used in the third stage of training.

3. The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic Distillation

中文 AI Summary

本文在受控多轮环境中系统研究了长程规划能力的获取与 Post-Training 塑形（对比了 GRPO 与 OPD）。作者提出了多教师在线策略蒸馏（MOPD）来整合跨环境规划能力。基于摘要判断，研究揭示了 GRPO 与 OPD 在低质量/长程数据下的更新方向差异。

Abstract

Multi-turn long-horizon planning is critical for foundation model agents, yet how to fundamentally improve it remains unclear. Existing models are trained on uncontrollable and opaque Internet data, making it difficult to identify how planning ability is acquired, shaped, and integrated. To address this challenge, we introduce a unified and controlled multi-turn environment that enables precise control. It allows systematically study long-horizon planning across three stages. (1) Planning ability acquisition during pre-training. We study data format, distribution, and quality. Explicit world model construction through CoT state transition modeling yields stronger long-horizon generalization. Atomic skills alone are insufficient for compositional generalization, whereas a litte long-horizon data works. Moreover, suboptimal trajectories severely impair performance because errors amplify over long horizons. (2) Planning ability shaping via GRPO and OPD post-training. Through mutual information, we distinguish general planning patterns from task-specific planning knowledge. For planning patterns, we identify three application regions of post-training: unnecessary, effective, and unsupported. OPD has a broader effective region than GRPO under low-quality and long-horizon settings, as it provides more consistent update directions. For planning knowledge, distilling unseen procedures from a teacher with different knowledge may impair student’s prior world modeling without fully establishing new knowledge. (3) Planning ability integration through MOPD post-training. We show that multi-teacher on-policy distillation (MOPD) integrates capabilities by converging to shared planning-pattern across environments. Compatible patterns enable cross-environment generalization, partially shared patterns support continual learning, while completely conflicting patterns cause severe interference.

排序依据与全文待核验点

  • 为什么相关： Investigates post-training (GRPO, OPD, and Multi-Teacher On-Policy Agentic Distillation) for long-horizon planning patterns and world modeling.
  • 全文待核验： Verify how GRPO and OPD are specifically adapted for multi-turn planning trajectories and mutual information analysis.
  • 方向： RLHF-VLM/LLM
  • 研究阅读价值： 3/5
  • 潜在工程价值： 4/5
  • 作者： Junlin Liu, Jiangwang Chen, Zixin Song, Shuaiyu Zhou, Chunji Lv, Hank Wu, Kailin Jiang, Jinyang Wu, Bohan Yu, Chenxi Zhou
  • Hugging Face： 论文页
  • arXiv： 摘要页 · PDF
  • 代码： 未提供
  • 项目页： 未提供

中文 AI Summary

针对 Agent 搜索中结果导向 RL 监督信号稀疏的问题，本文提出 Multi-Agent Protocol Distillation (MAPD) 框架。通过提取 JSON 协议作为特权分支的密集蒸馏信号，并与稀疏 RL 目标共同训练。基于摘要判断，该方法有效防止了策略漂移和冗长退化。

Abstract

Agentic search enables large language models to solve knowledge-intensive tasks by interleaving multi-step reasoning with retrieval, yet optimizing this with outcome-based reinforcement learning (RL) provides only sparse supervision. Knowledge distillation can supply denser guidance, and advanced proprietary models with their strong reasoning capabilities are promising teachers. While distilling from proprietary models can densify this supervisory signal, conventional logit-matching is precluded by hidden logits and mismatched tokenizers, whereas raw natural language trajectory imitation transfers superficial stylistic artifacts rather than core reasoning competence. To address the heterogeneous distillation problem and bridge the distribution gap, we propose Multi-Agent Protocol Distillation (MAPD), a joint distillation and RL framework uses a structured, style-normalized protocol as an intermediate representation. An offline multi-agent system (MAS) decomposes each query, retrieves supporting evidence, repairs failed searches, and converts the resulting exploration trace into a JSON protocol containing the task type, reasoning plan, and extractive grounding facts. During training, the protocol is provided only to a privileged branch of the student policy, whose token distributions furnish a dense distillation signal alongside the sparse RL objective. Extensive evaluations across seven QA benchmarks demonstrate that MAPD consistently outperforms competitive distillation and RL, achieving average success rates of 39.4% on Qwen3-1.7B and 44.4% on Qwen3-4B. Crucially, the framework generalizes robustly across diverse proprietary teachers while effectively mitigating the student policy from style drift and verbosity degeneration.

排序依据与全文待核验点

  • 为什么相关： Combines reinforcement learning (sparse RL objective) with multi-agent protocol distillation for agentic search and reasoning.
  • 全文待核验： Verify the implementation details of the joint loss combining the sparse RL objective and the dense protocol distillation loss.

5. Kimi K3: Open Frontier Intelligence

  • 方向： RLHF-VLM/LLM
  • 研究阅读价值： 3/5
  • 潜在工程价值： 4/5
  • 作者： Kimi Team, Tongtong Bai, Yifan Bai, Yiping Bao, M. C., Jianfeng Cai, Xinyuan Cai, Peizhou Cao, Yuxuan Cao, Ziwei Chai, Y. Charles, H. S. Che, Guanduo Chen, Guangyu Chen, Guanzheng Chen, Huarong Chen, Jia Chen, Jianlong Chen, Jun Chen, Kexin Chen, Peng Chen, Ruijue Chen, Wentao Chen, Xin Chen, Yang Chen, Yanru Chen, Yifei Chen, Yingjiang Chen, Yuankun Chen, Yujie Chen, Yutian Chen, Zhirong Chen, Dazhi Cheng, Yean Cheng, Jialei Cui, Jingbing Cui, Anqi Dai, Jiaqi Deng, Hao Ding, Rui Ding, Shaofeng Ding, Mengfan Dong, Mengnan Dong, Yuhao Dong, Yuxin Dong, Angang Du, Chenzhuang Du, Dikang Du, Jusen Du, Yulun Du, Yu Fan, Jing Feng, Qiulin Feng, Yichen Feng, Kelin Fu, Qiang Fu, Fuxuan Gao, Hongcheng Gao, Jingyue Gao, Tong Gao, Weijia Gao, Shangyi Geng, Jie Gong, Linhu Gong, Shengao Gong, Xiaochen Gong, Qizheng Gu, Yicheng Gu, Shuhao Guan, Haiqing Guo, Shiqi Guo, Xiang Guo, Zhengyan Guo, Beixi Hao, Wenxin Hao, Xiaoru Hao, Dailan He, Haotian He, Lehan He, Qi He, Weiran He, Xinran He, Xinyi He, Yibo He, Yunjia He, Chao Hong, Tiange Hong, Hao Hu, Jiaxi Hu, Ruikun Hu, Weiming Hu, Yangyang Hu, Zhenxing Hu, Liang Hua, Jinbin Huang, Ke Huang, Ruiyuan Huang, Siying Huang, Weixiao Huang, Yan Huang, Zhengjie Huang, Zhiqi Huang, Yulong Hui, Chaobo Jia, Yutong Jiang, Zhejun Jiang, Zuoyou Jiang, Wenyi Jin, Xinyi Jin, Yu Jing, Huanjun Kong, Guokun Lai, Aidi Li, Cheng Li, Chengyuan Li, Cong Li, Fang Li, Guanyu Li, Haoyang Li, Jia Li, Junxiong Li, Lei Li, Letian Li, Lincan Li, Weihong Li, Wentao Li, Xintong Li, Yang Li, Yishen Li, Yiwei Li, Yuxiao Li, Zhaowei Li, Zhaoxi Li, Zheming Li, Zhengxiao Li, Zhiyuan Li, Jiawei Lin, Xiaohan Lin, Yibo Lin, Zichao Lin, Ziyan Lin, Bill Liu, Boxiao Liu, Chuan Liu, Liang Liu, Shaowei Liu, Shudong Liu, Shuran Liu, Tianwei Liu, Weizhou Liu, Yangyang Liu, Yanming Liu, Yibo Liu, Yipeng Liu, Zhengying Liu, Zhiheng Liu, Enzhe Lu, Haoyu Lu, Linqiang Lu, Tingzhan Lu, Zhiyuan Lu, Aotian Luo, G. Luo, Junyu Luo, Yifan Luo, B. Lyu, Wenzhou Lyu, Shaoguang Mao, Yuan Mei, Xin Men, Minqing Ni, Yixuan Niu, Siyuan Pan, Shujun Peng, Zhangyang Qi, Ruoyu Qin, ZeChao Qin, Zeyu Qin, Haiquan Qiu, Jianxin Qiu, Jiezhong Qiu, Bowen Qu, Yuhao Qu, Zeyu Shang, Youbo Shao, Han Shen, Jincheng Shi, Juanfeng Shi, Lidong Shi, Shengyuan Shi, Wingchun Siu, Pengwei Song, Xiaoxi Song, Jianlin Su, Yunfeng Su, Zhaochen Su, Lin Sui, Jingsong Sun, Junyao Sun, Shaoning Sun, Shuzhe Sun, Tongyu Sun, Yujun Sun, Yunpeng Tai, Chuning Tang, Heyi Tang, Sirui Tang, Zecheng Tang, Chaoran Tian, Rongpeng Tian, Yu Tian, Wei Tu, Chensi Wang, Chuang Wang, Chunjie Wang, Dinglu Wang, Feng Wang, Hailong Wang, Haiming Wang, Hao Wang, Hao Wang, Huaqing Wang, Hui Wang, Jiayi Wang, Jinglong Wang, Jinhong Wang, Jiuzheng Wang, Linian Wang, Shaobo Wang, Shenzhi Wang, Shuyi Wang, Si Wang, Siyuan Wang, Tianfu Wang, Wenjue Wang, Xingran Wang, Xinmei Wang, Xinyuan Wang, Xusheng Wang, Yalin Wang, Yangkun Wang, Yao Wang, Yaoyu Wang, Yejie Wang, Yiqin Wang, Yucheng Wang, Yuzhi Wang, Zhaoji Wang, Zhaowei Wang, Zhengtao Wang, Zhenhao Wang, Zhongsheng Wang, Zifan Wang, Chu Wei, Ming Wei, Shouxin Wei, Zichen Wen, Fan Wu, Haoning Wu, Rucong Wu, Wenhao Wu, Xiaoxue Wu, Yingcong Wu, Yongqi Wu, Yuxin Wu, Zijian Wu, Xinglang Xian, Chenxuan Xiang, Yuye Xiang, Bocheng Xiao, Chenjun Xiao, Xin Xiao, Jin Xie, Xiaotong Xie, Yifeng Xie, Zhe Xie, Bowei Xing, Yiming Xiong, Baosheng Xu, Boyu Xu, Jiale Xu, Jianfan Xu, Jing Xu, Jinjing Xu, L. H. Xu, Qingtao Xu, Shuyao Xu, Suting Xu, Tiantian Xu, Tianxiang Xu, Weixin Xu, Xinran Xu, Yangchuan Xu, Ye Xu, Yueni Xu, Ziyao Xu, Haonan Xue, Junjie Yan, Yaoyao Yan, Fan Yang, Guangyao Yang, Hao Yang, Junwei Yang, Ruoyu Yang, Wenjie Yang, Xiaofei Yang, Xinyu Yang, Yi Yang, Yiling Yang, Ying Yang, Yuchen Yang, Zhen Yang, Zhilin Yang, Zian Yang, Zuhao Yang, Haotian Yao, Dan Ye, Haoran Ye, Wenjie Ye, Zhanbo Ye, Bohong Yin, Haoxiang Yin, Xietong Yin, Chengzhen Yu, Haozhen Yu, Longhui Yu, Shengnan Yu, Shuying Yu, Tianxiang Yu, Enming Yuan, Mengjie Yuan, Tongtian Yue, Wei Yue, Yang Yue, Dunyuan Zha, Haobing Zhan, B. H. Zhang, Dehao Zhang, Fei Zhang, Hao Zhang, Haoyuan Zhang, Huanyu Zhang, Jiapei Zhang, Jiaxuan Zhang, Jin Zhang, Kaiyi Zhang, Miaozhen Zhang, Puqi Zhang, Qinglei Zhang, Rong Zhang, Rui Zhang, Shaoshuai Zhang, Shiyi Zhang, Xiaobin Zhang, Xiaoyun Zhang, Y. Zhang, Yangkun Zhang, Ye Zhang, Yichi Zhang, Yikun Zhang, Yizhi Zhang, Yongting Zhang, Yu Zhang, Yutao Zhang, Yutong Zhang, Zheng Zhang, Zijing Zhang, Bin Zhao, Chenguang Zhao, Feifan Zhao, Jinglun Zhao, Jinxiang Zhao, Shuai Zhao, Wenshuo Zhao, Xiangyu Zhao, Xuanle Zhao, Yikai Zhao, Zijia Zhao, Haozhi Zheng, Huabin Zheng, Ruihan Zheng, Shaojie Zheng, Tengyang Zheng, Haofeng Zhong, Lei Zhong, Longguang Zhong, M. Zhou, Qiankang Zhou, Runjie Zhou, Ruozhang Zhou, Xinyu Zhou, Yiqiao Zhou, Zaida Zhou, Jinguo Zhu, Liya Zhu, Xinhao Zhu, Yangjunfeng Zhu, Yuxuan Zhu, Zhen Zhu, Chen Zhuang, Weiyu Zhuang, Xinxing Zu
  • Hugging Face： 论文页
  • arXiv： 摘要页 · PDF
  • 代码： https://github.com/MoonshotAI/Kimi-K3
  • 项目页： https://www.kimi.com/blog/kimi-k3

中文 AI Summary

本文介绍了开源 2.8T 参数 MoE 模型 Kimi K3，其后训练阶段强调跨通用、Agent 和代码领域的强化学习（RL），支持多推理工作量级别。基于摘要判断，模型实现了百万人程 Agentic RL 状态持久化与高效率扩展。

Abstract

We introduce Kimi K3, a 2.8T parameter Mixture-of-Experts model with 104 billion activated parameters, native vision capabilities, and a 1-million-token context window. Kimi K3 is built on Kimi Delta Attention and Attention Residuals, which improve information flow across sequence length and model depth. Together with Stable LatentMoE, which effectively activates 16 of 896 routed experts per token, and refined training and data recipes, these advances yield an approximately 2.5x improvement in overall scaling efficiency over Kimi K2. Post-training highlights reinforcement learning across general, agentic, and coding domains and multiple reasoning-effort levels, enabling compositional generalization and robust long-horizon execution. At 2.8T scale, Kimi K3 is supported by infrastructure advances in multiple areas: algorithm-system co-design for KDA, perfectly balanced expert-parallel training with efficient memory management, million-token agentic RL with persistent rollout and sandbox states, and deployment innovations. Extensive evaluations show that Kimi K3 achieves frontier-level performance across long-horizon coding, agentic, knowledge, reasoning, and vision tasks. While its overall performance still trails the most powerful proprietary models, namely Claude Fable 5 and GPT-5.6 Sol, Kimi K3 consistently outperforms other open and proprietary models evaluated in our suite. We release the full Kimi K3 model weights to facilitate future research and accelerate the broader deployment and adoption of frontier intelligence.

排序依据与全文待核验点

  • 为什么相关： Reports post-training reinforcement learning on a 2.8T MoE native VLM/LLM across general, agentic, and coding domains.
  • 全文待核验： Verify the specific RL algorithms, reward model signals, or PPO/GRPO configurations detailed in the post-training section.

筛选备注

Filtered out non-scoped papers, including general tool-use/computer-use harnesses, robotics/embodied manipulation papers (e.g. 2607.21655, 2607.24744, 2607.23909), bio/molecule models (2607.23518), and pure inference sparsification/eval benchmarks without RLHF or post-training alignment contribution.

Edit

继续完善这篇笔记

编辑本文AI 协助

关系图谱