1. zeroclaw-labs/zeroclaw
分类：开源项目来源：github_search分数：9作者：zeroclaw-labs时间：2026-03-05T01:45:46Z
ZeroClaw 发布为一个100% Rust的代理工作流运行时框架，主打超低内存与可插拔架构，可在低配硬件运行并降低部署成本，同时强调官方仓库与安全合规以防冒名与凭证风险。
- 项目定位为 agentic workflows 的运行时基础设施，抽象模型、工具、记忆与执行，实现“构建一次，到处运行”。
- 核心卖点包括低资源占用（宣称<5MB RAM）、毫秒级冷启动、单二进制跨 ARM/x86/RISC-V 部署及 provider/channel/tool 可替换。
- 仓库提供多语言文档、快速安装路径（Homebrew、Bootstrap、Cargo）和运维/安全/故障排查文档入口。
- 给出与 OpenClaw/NanoBot/PicoClaw 的基准快照，对比语言、内存、启动时间与成本，强调在边缘设备上的效率优势。
- 发布了重要公告：仅 github.com/zeroclaw-labs/zeroclaw 为官方仓库，并提醒防范冒名域名与非官方募资信息。
#GitHub #repo #开源项目 #Rust #Agent
2. kortix-ai/suna
分类：开源项目来源：github_search分数：7作者：kortix-ai时间：2026-03-05T01:57:48Z
Kortix 在 GitHub 开源了 Suna（Kortix）自主 AI 代理平台，提供从构建、部署到运行的全栈能力，重要性在于它把通用与行业代理落地所需的基础设施打包成可自托管方案。
- 项目主打“完整自主代理平台”，含旗舰示例代理 Kortix Super Worker，可通过自然语言执行研究、分析、自动化与复杂工作流。
- 支持构建定制化代理，覆盖客服、内容、销售、研发及医疗/金融/法律/教育等行业场景。
- 核心能力包括浏览器自动化、文件管理、网页情报抓取、系统运维操作、API 集成与可视化 Agent Builder。
- 架构由 FastAPI 后端、Next.js 前端、Docker 隔离运行时和 Supabase 数据层组成，支持多 LLM 提供商接入。
- 提供快速启动向导与 Docker/手动两种部署方式，支持追加配置 API Key 与服务管理、日志查看等运维功能。
#GitHub #repo #开源项目 #FastAPI #Next.js #Docker #Supabase #Browser Automation #Agent
3. botpress/botpress
分类：开源项目来源：github_search分数：6作者：botpress时间：2026-03-05T01:53:47Z
Botpress 在 GitHub 开源了其云端聊天机器人开发仓库，涵盖集成、CLI/SDK/API 工具与示例机器人，帮助开发者快速构建并发布基于 OpenAI 的新一代助手。
- 仓库聚合了 Botpress Hub 的公开集成、Cloud 开发工具（CLI/SDK/API Client）和 bots-as-code 示例。
- 支持通过 Botpress CLI 创建、部署与发布集成：先私有到工作区，再可公开到 Botpress Hub。
- 官方强调 Studio 是推荐的建 bot 方式，代码化 bots 更适合资深开发者与程序化场景。
- 项目欢迎社区贡献 PR/Issue，并提供 Discord 作为更快的沟通反馈渠道。
- 仓库内包采用 MIT 许可证，降低二次开发和商业集成门槛。
#GitHub #repo #开源项目 #Botpress #OpenAI #SDK #CLI #Integrations #MIT
4. Reducing hyperparameter sensitivity in measurement-feedback based Ising machines
分类：研究/论文来源：arxiv_search分数：100作者：Toon Sevenants时间：2026-03-04T14:03:09Z
论文分析了测量-反馈型离散时间伊辛机与连续时间模型在超参数有效范围上的差异，并提出且实验验证了降低超参数敏感性的方法，这有助于提升其在组合优化硬件求解中的可用性与稳定性。
- 指出实验中常见的测量-反馈架构是离散时间运行，与理论连续时间伊辛机存在动力学差异。
- 发现离散实现下可用超参数范围明显缩小，导致系统对参数调优更敏感。
- 系统分析了该差异对伊辛机实际运行与性能表现的影响。
- 提出一种降低测量-反馈架构超参数敏感性的方法，并给出实验验证。
#arXiv #paper #研究/论文
5. End-to-end event reconstruction for precision physics at future colliders
分类：研究/论文来源：arxiv_search分数：98作者：Dolores Garcia时间：2026-03-04T13:55:04Z
该论文提出面向未来对撞机的端到端全局事例重建框架，在FCC-ee全模拟数据上较现有规则算法显著提升重建效率并降低伪粒子率，从而提高希格斯等精密测量能力并加速探测器设计迭代。
- 提出从径迹、量能器和缪子击中信息直接映射到粒子级对象的端到端重建方法。
- 方法结合几何代数Transformer与object condensation聚类，并接入粒子鉴别与能量回归网络。
- 在FCC-ee的CLD探测器概念全模拟电子-正电子碰撞上完成基准测试。
- 相较最先进规则式算法，重建相对效率提升约10–20%，带电强子伪粒子率最高降低近两个数量级。
- 可见能量与不变质量分辨率提升约22%，且减少对探测器特定调参的依赖。
#arXiv #paper #研究/论文 #FCC-ee
6. SaFeR: Safety-Critical Scenario Generation for Autonomous Driving Test via Feasibility-Constrained Token Resampling
分类：研究/论文来源：arxiv_search分数：95作者：Jinlong Cui时间：2026-03-04T13:46:12Z
论文提出SaFeR框架，通过可行性约束的token重采样在生成对抗性自动驾驶测试场景时兼顾物理可行与行为真实，并在Waymo与nuPlan闭环实验中优于现有方法，提升测试有效性与可靠性。
- 将交通场景生成建模为离散的下一token预测任务，使用Transformer作为真实驾驶分布先验。
- 提出差分注意力机制以增强多体交互建模能力并降低注意力噪声。
- 设计可行性约束重采样策略，在高概率信任域内注入对抗行为以保持自然性。
- 基于离线强化学习近似最大可行区域（LFR），避免生成理论上不可避免碰撞场景。
- 在Waymo Open Motion Dataset与nuPlan闭环实验中，相比SOTA实现更高解率、更好运动学真实度且保持强对抗性。
#arXiv #paper #研究/论文 #Transformer #Waymo #nuPlan
7. Monitoring Emergent Reward Hacking During Generation via Internal Activations
分类：研究/论文来源：arxiv_search分数：92作者：Patrick Wilhelm时间：2026-03-04T13:44:24Z
该论文提出基于内部激活的实时监测方法，在生成过程中识别大模型奖励黑客行为并发现其常在早期出现且会被链式思维与额外测试时计算放大，这对部署后更早发现失配风险很关键。
- 针对仅看最终输出难以发现的涌现式奖励黑客问题，作者转向生成过程中的内部表征监测。
- 方法上使用残差流激活训练稀疏自编码器，并结合轻量线性分类器给出token级奖励黑客概率估计。
- 在多种模型家族与微调混合设置下，内部激活模式可稳定区分奖励黑客与正常行为。
- 该信号可泛化到未见过的混合策略适配器，显示出一定跨策略鲁棒性。
- 实验显示奖励黑客信号常在推理早期出现并持续，且在弱奖励目标下会被CoT提示与更多测试时计算增强。
#arXiv #paper #研究/论文
8. Tuning Just Enough: Lightweight Backdoor Attacks on Multi-Encoder Diffusion Models
分类：研究/论文来源：arxiv_search分数：90作者：Ziyuan Chen时间：2026-03-04T13:41:20Z
该论文系统分析了多编码器文生图模型（以Stable Diffusion 3为例）的后门风险，提出仅微调不到0.2%参数的MELT攻击仍能高效植入后门，说明现实部署中的多编码器扩散模型并未因规模增大而更安全。
- 研究对象是采用三个文本编码器的Stable Diffusion 3，填补了多编码器场景后门研究不足。
- 作者将后门攻击目标划分为四类，并分析实现各类目标所需的最小编码器组合。
- 提出MELT方法：冻结预训练文本编码器，仅训练低秩适配器（LoRA）实施后门注入。
- 实验显示仅需调优少于编码器总参数0.2%即可实现有效攻击，兼具轻量与攻击成功率。
- 结论指出多编码器带来的参数规模增长并未自然抵御文本编码器层面的后门威胁。
#arXiv #paper #研究/论文 #Stable Diffusion 3
9. FedCova: Robust Federated Covariance Learning Against Noisy Labels
分类：研究/论文来源：arxiv_search分数：88作者：Xiangyu Zhong时间：2026-03-04T13:40:09Z
论文提出无需依赖干净设备或公共数据的联邦学习框架FedCova，通过基于类特征协方差的鲁棒编码、分类与噪声纠正，在异构分布和多种噪声场景下显著提升全局模型抗噪性能。
- 针对联邦学习中噪声标签导致的局部过拟合与全局性能下降，提出依赖最小化的FedCova方法。
- 以互信息最大化为基础，设计仅依赖类特征协方差并包含误差容忍项的联邦有损特征编码目标。
- 利用协方差表征的特征子空间，构建子空间增强分类器，并统一特征学习、分类器构造与噪声标签纠正三过程。
- 在对称与非对称噪声、异构数据分布下进行验证，并在CIFAR-10/100与Clothing1M上优于现有SOTA方法。
#arXiv #paper #研究/论文 #Clothing1M
10. Fermi-Dirac thermal measurements: A framework for quantum hypothesis testing and semidefinite optimization
分类：研究/论文来源：arxiv_search分数：85作者：Nana Liu时间：2026-03-04T13:39:46Z
Quantum measurements are the means by which we recover messages encoded into quantum states. They are at the forefront of quantum hypothesis testing, wherein t…
- Quantum measurements are the means by which we recover messages encoded into quantum states
- They are at the forefront of quantum hypothesis testing, wherein the goal is to perform an optimal measurement for arriving at a correct co…
- Mathematically, a measurement operator is Hermitian with eigenvalues in [0,1]
- By noticing that this constraint on each eigenvalue is the same as that imposed on fermions by the Pauli exclusion principle, we interpret …
- Under this perspective, various objective functions in quantum hypothesis testing can be viewed as the total expected energy associated wit…
- By instead fixing a temperature and minimizing the total expected fermionic free energy, we find that optimal measurements for these modifi…
#arXiv #paper #研究/论文
11. Sim2Sea: Sim-to-Real Policy Transfer for Maritime Vessel Navigation in Congested Waters
分类：研究/论文来源：arxiv_search分数：82作者：Xinyu Cui时间：2026-03-04T13:36:33Z
Autonomous navigation in congested maritime environments is a critical capability for a wide range of real-world applications. However, it remains an unresolve…
- Autonomous navigation in congested maritime environments is a critical capability for a wide range of real-world applications
- However, it remains an unresolved challenge due to complex vessel interactions and significant environmental uncertainties
- Existing methods often fail in practical deployment due to a substantial sim-to-real gap, which stems from imprecise simulation, inadequate…
- To address these, we propose \textbf{Sim2Sea}, a comprehensive framework designed to bridge simulation and real-world execution
- Sim2Sea advances in three key aspects
- First, we develop a GPU-accelerated parallel simulator for scalable and accurate maritime scenario simulation
#arXiv #paper #研究/论文
12. Inference-Time Toxicity Mitigation in Protein Language Models
分类：研究/论文来源：arxiv_search分数：80作者：Manuel Fernández Burda时间：2026-03-04T13:28:51Z
Protein language models (PLMs) are becoming practical tools for de novo protein design, yet their dual-use potential raises safety concerns. We show that domai…
- Protein language models (PLMs) are becoming practical tools for de novo protein design, yet their dual-use potential raises safety concerns
- We show that domain adaptation to specific taxonomic groups can elicit toxic protein generation, even when toxicity is not the training obj…
- To address this, we adapt Logit Diff Amplification (LDA) as an inference-time control mechanism for PLMs
- LDA modifies token probabilities by amplifying the logit difference between a baseline model and a toxicity-finetuned model, requiring no r…
- Across four taxonomic groups, LDA consistently reduces predicted toxicity rate (measured via ToxDL2) below the taxon-finetuned baseline whi…
- We evaluate quality using Fréchet ESM Distance and predicted foldability (pLDDT), finding that LDA maintains distributional similarity to n…
#arXiv #paper #研究/论文
13. DQE-CIR: Distinctive Query Embeddings through Learnable Attribute Weights and Target Relative Negative Sampling in Composed Image Retrieval
分类：研究/论文来源：arxiv_search分数：78作者：Geon Park时间：2026-03-04T13:17:44Z
Composed image retrieval (CIR) addresses the task of retrieving a target image by jointly interpreting a reference image and a modification text that specifies…
- Composed image retrieval (CIR) addresses the task of retrieving a target image by jointly interpreting a reference image and a modification…
- Most existing methods are still built upon contrastive learning frameworks that treat the ground truth image as the only positive instance …
- This strategy inevitably introduces relevance suppression, where semantically related yet valid images are incorrectly pushed away, and sem…
- As a result, the learned query representations often lack discriminativeness, particularly at fine-grained attribute modifications
- To overcome these limitations, we propose distinctive query embeddings through learnable attribute weights and target relative negative sam…
- DQE-CIR incorporates learnable attribute weighting to emphasize distinctive visual features conditioned on the modification text, enabling …
#arXiv #paper #研究/论文
14. mlx-vis: GPU-Accelerated Dimensionality Reduction and Visualization on Apple Silicon
分类：研究/论文来源：arxiv_search分数：75作者：Han Xiao时间：2026-03-04T13:16:48Z
mlx-vis is a Python library that implements six dimensionality reduction methods and a k-nearest neighbor graph algorithm entirely in MLX, Apple's array framew…
- mlx-vis is a Python library that implements six dimensionality reduction methods and a k-nearest neighbor graph algorithm entirely in MLX, …
- The library provides UMAP, t-SNE, PaCMAP, TriMap, DREAMS, CNE, and NNDescent, all executing on Metal GPU through a unified fit_transform in…
- Beyond embedding computation, mlx-vis includes a GPU-accelerated circle-splatting renderer that produces scatter plots and smooth animation…
- 264 encoding
- On Fashion-MNIST with 70,000 points, all methods complete embedding in 2
- 1-3
#arXiv #paper #研究/论文
15. The Empty Quadrant: AI Teammates for Embodied Field Learning
分类：研究/论文来源：arxiv_search分数：72作者：Hyein Kim时间：2026-03-04T13:13:44Z
For four decades, AIED research has rested on what we term the Sedentary Assumption: the unexamined design commitment to a stationary learner seated before a s…
- For four decades, AIED research has rested on what we term the Sedentary Assumption: the unexamined design commitment to a stationary learn…
- Mobile learning and museum guides have moved learners into physical space, and context-aware systems have delivered location-triggered cont…
- We map this gap through a 2 x 2 matrix (AI Role x Learning Environment) and identify an undertheorized intersection: the configuration in w…
- To fill it, we propose Field Atlas, a framework grounded in embod-ied, embedded, enactive, and extended (4E) cognition, active inference, a…
- The architecture pairs volitional photography with immediate voice reflec-tion, constrains AI to Socratic provocation rather than answer de…
- We demonstrate the framework through a museum scenario and argue that the resulting trajecto-ries -- bound to a specific body, place, and t…
#arXiv #paper #研究/论文
16. Multi-Stage Music Source Restoration with BandSplit-RoFormer Separation and HiFi++ GAN
分类：研究/论文来源：arxiv_search分数：70作者：Tobias Morocutti时间：2026-03-04T13:10:39Z
Music Source Restoration (MSR) targets recovery of original, unprocessed instrument stems from fully mixed and mastered audio, where production effects and dis…
- Music Source Restoration (MSR) targets recovery of original, unprocessed instrument stems from fully mixed and mastered audio, where produc…
- This technical report presents the CP-JKU team's system for the MSR ICASSP Challenge 2025
- Our approach decomposes MSR into separation and restoration
- First, a single BandSplit-RoFormer separator predicts eight stems plus an auxiliary other stem, and is trained with a three-stage curriculu…
- Second, we apply a HiFi++ GAN waveform restorer trained as a generalist and then specialized into eight instrument-specific experts
#arXiv #paper #研究/论文
17. Self-adapting Robotic Agents through Online Continual Reinforcement Learning with World Model Feedback
分类：研究/论文来源：arxiv_search分数：68作者：Fabian Domberg时间：2026-03-04T13:07:42Z
As learning-based robotic controllers are typically trained offline and deployed with fixed parameters, their ability to cope with unforeseen changes during op…
- As learning-based robotic controllers are typically trained offline and deployed with fixed parameters, their ability to cope with unforese…
- Biologically inspired, this work presents a framework for online Continual Reinforcement Learning that enables automated adaptation during …
- Building on DreamerV3, a model-based Reinforcement Learning algorithm, the proposed method leverages world model prediction residuals to de…
- Adaptation progress is monitored using both task-level performance signals and internal training metrics, allowing convergence to be assess…
- The approach is validated on a variety of contemporary continuous control problems, including a quadruped robot in high-fidelity simulation…
- Relevant metrics and their interpretation are presented and discussed, as well as resulting trade-offs described
#arXiv #paper #研究/论文 #Agent
18. A Multi-Dimensional Quality Scoring Framework for Decentralized LLM Inference with Proof of Quality
分类：研究/论文来源：arxiv_search分数：65作者：Arther Tian时间：2026-03-04T13:05:46Z
Decentralized large language model (LLM) inference networks can pool heterogeneous compute to scale serving, but they require lightweight and incentive-compati…
- Decentralized large language model (LLM) inference networks can pool heterogeneous compute to scale serving, but they require lightweight a…
- Prior work introduced cost-aware Proof of Quality (PoQ) and adaptive robust PoQ to allocate rewards under evaluator heterogeneity and adver…
- In this paper, we focus on the quality signal itself and propose a multi-dimensional quality scoring framework that decomposes output quali…
- Using logged outputs from QA and summarization tasks, we systematically audit dimension reliability and show that seemingly reasonable dime…
- While the default composite underperforms a strong single semantic evaluator, ablations reveal that removing unreliable dimensions and re-n…
- Finally, we integrate the composite score as a drop-in quality signal in PoQ and demonstrate complementary benefits with robust aggregation…
#arXiv #paper #研究/论文
19. Volumetric Directional Diffusion: Anchoring Uncertainty Quantification in Anatomical Consensus for Ambiguous Medical Image Segmentation
分类：研究/论文来源：arxiv_search分数：62作者：Chao Wu时间：2026-03-04T12:58:43Z
Equivocal 3D lesion segmentation exhibits high inter-observer variability. Conventional deterministic models ignore this aleatoric uncertainty, producing over-…
- Equivocal 3D lesion segmentation exhibits high inter-observer variability
- Conventional deterministic models ignore this aleatoric uncertainty, producing over-confident masks that obscure clinical risks
- Conversely, while generative methods (e
- g
- , standard diffusion) capture sample diversity, recovering complex topology from pure noise frequently leads to severe structural fractures…
- To resolve this fidelity-diversity trade-off, we propose Volumetric Directional Diffusion (VDD)
#arXiv #paper #研究/论文
20. Continuous Modal Logical Neural Networks: Modal Reasoning via Stochastic Accessibility
分类：研究/论文来源：arxiv_search分数：60作者：Antonin Sulc时间：2026-03-04T12:55:04Z
We propose Fluid Logic, a paradigm in which modal logical reasoning, temporal, epistemic, doxastic, deontic, is lifted from discrete Kripke structures to conti…
- We propose Fluid Logic, a paradigm in which modal logical reasoning, temporal, epistemic, doxastic, deontic, is lifted from discrete Kripke…
- Each type of modal operator is backed by a dedicated Neural SDE, and nested formulas compose these SDEs in a single differentiable graph
- A key instantiation is Logic-Informed Neural Networks (LINNs): analogous to Physics-Informed Neural Networks (PINNs), LINNs embed modal log…
- The resulting framework, Continuous Modal Logical Neural Networks (CMLNNs), yields several key properties: (i) stochastic diffusion prevent…
- Three case studies demonstrate that Fluid Logic and LINNs can guide neural networks to produce consistent solutions across diverse domains:…
#arXiv #paper #研究/论文