01
AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security
Dongrui Liu, Yu Li, Zhonghao Yang, Peng Wang, Guanxu Chen, Yuejin Xie 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, rl, code, alignment, safety, dataset, training；Total 21.0/25。
- 方法要点
- Modern open-world agents such as OpenClaw exhibit powerful cross-environment execution capabilities yet introduce broad new safety risk sources. Meanwhile, advanced frontier AI models drastically lower attack barriers, rendering current agent alignment frameworks inadequate for real-world deployment. To tackle these emerging thr…
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.9Substance 3.9Evidence 4.0Actionability 4.2
02
YoCausal: How Far is Video Generation from World Model? A Causality Perspective
You-Zhe Xie, Yu-Hsuan Li, Jie-Ying Lee, Kaipeng Zhang, Yu-Lun Liu, Zhixiang Wang · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇围绕推理/规划能力给出方法或评测信号，适合纳入 Lucian 的 agent 研究路线。
- 为什么重要
- 命中：reasoning, rag, rl, evaluation, benchmark, dataset, diffusion；Total 20.6/25。
- 方法要点
- As video diffusion models (VDMs) advance toward world models, a key question arises: do they truly understand causality, or merely overfit to statistical temporal patterns? Existing benchmarks mostly rely on synthetic data, limiting real-world generalization due to the sim-to-real gap. We present YoCausal, a two-level benchmark …
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 4.6Novelty 3.9Substance 3.9Evidence 4.0Actionability 4.2
03
Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments
Qiuyue Wang, Mingsheng Li, Jian Guan, Jinhui Ye, Sicheng Xie, Yitao Liu 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇围绕推理/规划能力给出方法或评测信号，适合纳入 Lucian 的 agent 研究路线。
- 为什么重要
- 命中：reasoning, rag, rl, benchmark, code, language model, training, vision-language；Total 20.4/25。
- 方法要点
- Embodied intelligence is often studied through specialized models for individual tasks such as manipulation or navigation, resulting in fragmented capabilities and limited generalization across tasks, environments, and robot embodiments. In this work, we study whether heterogeneous embodied decision-making problems can be unifie…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.9Substance 3.9Evidence 3.4Actionability 4.2
04
AsyncTool: Evaluating the Asynchronous Function Calling Capability under Multi-Task Scenarios
Kou Shi, Ziao Zhang, Shiting Huang, Avery Nie, Zhen Fang, Qiuchen Wang 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, tool, tools, reasoning, feedback, rl, evaluation；Total 20.4/25。
- 方法要点
- Large language model (LLM)-based agents have shown strong capabilities in using external tools to solve complex tasks. However, existing evaluations often overlook the temporal dimension of tool use, especially the impact of tool response latency, and are usually limited to single-task settings. In real-world applications, multi…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.9Substance 3.9Evidence 3.4Actionability 4.2
05
WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction
Chengzhi Liu, Yuzhe Yang, Sophia Xiao Pu, Yepeng Liu, Lin Long, Yichen Guo 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, memory, long-horizon, feedback, rag, retrieval, multimodal；Total 20.4/25。
- 方法要点
- Multimodal large language models are increasingly deployed as long-horizon agents, where memory must do more than recall: it must track an evolving world, revise what has gone stale, and surface the right evidence at decision time. Existing benchmarks measure recall over static dialogue, collapse memory into a single end-of-task…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.9Substance 3.9Evidence 3.4Actionability 4.2
06
Skill0.5: Joint Skill Internalization and Utilization for Out-of-Distribution Generalization in Agentic Reinforcement Learning
Jiapeng Zhu, Jianxiang Yu, Yibo Zhao, Chengcheng Han, Qi Gu, Xunliang Cai 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, memory, reinforcement, rl, language model, large language；Total 20.4/25。
- 方法要点
- Equipping large language models with explicit skills has emerged as a promising paradigm for enabling autonomous agents to solve complex tasks. Agent skills can be inherently divided into general skills for broad cognitive transfer and task-specific skills for dynamic execution. However, existing skill-based reinforcement learni…
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.3Substance 3.9Evidence 4.0Actionability 4.2
07
DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation
Jusuk Lee, Seungjae Lee, Jonghun Shin, Hoseong Jung, Sungha Kim, Daesol Cho 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇与检索增强和知识使用有关，对构建可落地的智能体记忆/知识模块有启发。
- 为什么重要
- 命中：rag, multimodal, rl, code, alignment, training, vision-language；Total 19.7/25。
- 方法要点
- Robot manipulation critically depends on perception that preserves the action-relevant aspects of a scene. Yet most robot learning pipelines are built upon visual encoders pre-trained for static recognition or vision-language alignment, leaving motion understanding to downstream policies. We introduce DynaFLIP, a dynamics-aware …
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 4.2Novelty 3.3Substance 3.9Evidence 4.0Actionability 4.2
08
Multi-view Consistent 3D Gaussian Head Avatars 'without' Multi-view Generation
Aviral Chharia, Fernando De la Torre · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇提供评测或基准视角，有助于把想法转化为可比较的实验。
- 为什么重要
- 命中：rl, evaluation, code, alignment, dataset, training；Total 19.6/25。
- 方法要点
- High-fidelity 3D Gaussian head avatar generation is critical for applications such as AR/VR, telepresence, and digital humans. Existing methods depend on multi-view datasets, 3D captures, or intermediate 2D view synthesis. In contrast, we learn both conditional and unconditional 3D head models from randomly sampled 2D images alo…
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 3.5Novelty 3.9Substance 3.9Evidence 4.0Actionability 4.2
09
How LoRA Remembers? A Parametric Memory Law for LLM Finetuning
Ziwen Xu, Haiwen Hong, Linsong Yu, Benglei Cui, Longtao Huang, Hui Xue 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇提供评测或基准视角，有助于把想法转化为可比较的实验。
- 为什么重要
- 命中：memory, rl, evaluation, coding, code, llm, language model, large language；Total 19.3/25。
- 方法要点
- Large Language Models (LLMs) must continuously learn and update knowledge to remain effective in dynamic real-world environments. While Low-Rank Adaptation (LoRA) is widely used for such memory updates, existing studies mainly rely on qualitative downstream evaluations, leaving the quantitative capacity limits and underlying dyn…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 2.8Substance 3.9Evidence 3.4Actionability 4.2
10
When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systems
Corrado Rainone, Davide Belli, Bence Major, Arash Behboodi · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, rl, llm, language model, large language, inference；Total 19.2/25。
- 方法要点
- The design space of agentic AI inference spans two extremes: frontier large language models (LLMs), typically hosted in the cloud and offering strong performance across a wide range of tasks at substantially high cost, and more cost-efficient small language models (SLMs), which are amenable to on-device inference. Hybrid multi-a…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.3Substance 3.9Evidence 3.4Actionability 3.6
11
CoHyDE: Iterative Co-Training of LLM Rewriter & Dense Encoder for Tool Retrieval
Vaishali Senthil, Ashutosh Hathidara, Sebastian Schreiber · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇直接触及 agent/tool-use 工作流，可作为长程任务系统设计或评测的参考。
- 为什么重要
- 命中：agent, agents, tool, retrieval, code, llm, training；Total 19.0/25。
- 方法要点
- Tool retrieval over large API catalogs is a core bottleneck for LLM agents: user queries arrive in colloquial, often underspecified language, while the catalog uses technical API vocabulary that no fixed encoder can bridge on its own. The two dominant training approaches, contrastive encoder fine-tuning and HyDE-style query expa…
- 证据
- Evidence 2.6/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 3.3Substance 3.9Evidence 2.6Actionability 4.2
12
MoZoo:Unleashing Video Diffusion power in animal fur and muscle simulation
Dongxia Liu, Jie Ma, Xiaochen Yang, Jiancheng Zhang, Bin Xia, Zhehan Kan 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇提供评测或基准视角，有助于把想法转化为可比较的实验。
- 为什么重要
- 命中：rag, multimodal, benchmark, alignment, dataset, training, diffusion；Total 18.2/25。
- 方法要点
- The creation of cinematic-quality animal effects necessitates the precise modeling of muscle and fur dynamics, a process that remains both labor-intensive and computationally expensive within traditional production workflows. While generative diffusion models have shown promise in diverse artistic workflows, their capacity for h…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 4.0Novelty 2.8Substance 3.9Evidence 3.4Actionability 4.2
13
ChildVox: A Speech, Audio, and Large Audio-Language Model Benchmark in Understanding and Characterizing Sound across Childhood
Tiantian Feng, Anfeng Xu, Xuan Shi, Aditya Kommineni, Shakhrul Iman Siam, Megan Micheletti 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇提供评测或基准视角，有助于把想法转化为可比较的实验。
- 为什么重要
- 命中：benchmark, language model, dataset, speech；Total 18.2/25。
- 方法要点
- We present ChildVox, a novel benchmark for characterizing the diverse acoustic signals through which children communicate. Specifically, ChildVox follows the full developmental trajectory from birth through school age, covering physiological sounds, non-linguistic vocalizations, canonical syllables, and spoken language. ChildVox…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 2.9Novelty 3.9Substance 3.9Evidence 3.4Actionability 4.2
14
Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases
Dongyoon Hahm, Dylan Hadfield-Menell, Kimin Lee · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇在模型方法、训练或应用基础设施上有一定实质增量，值得快速浏览。
- 为什么重要
- 命中：feedback, reinforcement, rl, llm, language model, large language, alignment, dataset；Total 17.5/25。
- 方法要点
- Reinforcement Learning from Human Feedback (RLHF) is the standard method to align Large Language Models (LLMs) with human preferences. In this work, we introduce alignment tampering, a potential vulnerability where the LLM undergoing alignment influences the preference dataset, causing RLHF to amplify undesired behaviors. This a…
- 证据
- Evidence 3.4/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 5Novelty 2.8Substance 3.9Evidence 3.4Actionability 2.4
15
Towards Consistent Video Geometry Estimation
Zhu Yu, Jingnan Gao, Runmin Zhang, Lingteng Qiu, Zhengyi Zhao, Rui Peng 等 · Hugging Face Daily Papers 2026-05-31 + arXiv metadata
这篇扩展多模态模型能力，对未来多模态 agent 的输入/反馈闭环有参考价值。
- 为什么重要
- 命中：inference, dataset, training；Total 16.2/25。
- 方法要点
- This work presents ViGeo, a feed-forward foundation model for recovering spatially dense and temporally consistent geometry from video sequences. Built upon a plain transformer architecture without task-specific architectural modifications, ViGeo supports streaming, full-sequence, and long-video inference within a unified model.…
- 证据
- Evidence 4.0/5；需阅读全文核验基准与 ablation。
- 局限
- 仅基于公开摘要/元数据快速筛选；需要阅读全文确认实验设置、数据泄漏风险与复现细节。
- 下一步
- 阅读全文，提取可复现 baseline/指标；若与 agent、RAG、评测或个性化相关，可加入下一轮原型/benchmark backlog。
Relevance 2.0Novelty 3.9Substance 3.9Evidence 4.0Actionability 2.4