周报

周报

2026-W33 周报精选：The Defender’s Window；AiToEarn：Let's use AI to Ear；OpenAI joins PORTS-Pike proj

条目
80
覆盖天数
7
当前批次
2026-W33
往期
#08

豆包更新了工作模式，正面硬刚 Workbuddy

AI 办公场景现在成了兵家必争之地。 字节终于出手，豆包客户端最近更新，增加了全新的工作任务模式，体验跟 Codex 非常接近。 能用浏览器、能调 Skill，内置 Office 和飞书办公套件，同时支持本地和云端电脑，合上笔记本也能继续工作。 测试了 7-8 个任务，竟然有点超预期，宇宙厂终于在办公 Agent 场景

推荐理由AI 办公场景现在成了兵家必争之地。
创作与自媒体乔木博客
blog.qiaomu.ai打开原文
#14

Can We Defend Against AI-Generated Video Attacks on Real-World Crisis Events? A Systematic Evaluation of Detectors, Generators and Social Di

Recent video generators can fabricate realistic depictions of wars, disasters, public emergencies, and other real-world crises, creating substantial risks of misinformation. Existing benchmarks, however, provide limited

推荐理由Can We Defend Against AI-Generated：Recent video generators can fabricate realistic depictions of wars, disasters, public。
研究与学习HF daily papers
huggingface.co打开原文
#16

Apodex Discovery: Reality Benchmarks and Environments for Evaluating and Building Discoverative Artificial Intelligence

Apollo did not reach the Moon merely because its engineers could solve difficult equations. It succeeded by turning a distant ambition into a mission architecture of explicit objectives, simulation, verification, and rep

推荐理由Apodex Discovery: Reality Benchmarks：Apollo did not reach the Moon merely because its engineers could solve difficult。
研究与学习HF daily papers
huggingface.co打开原文
#17

Beyond Final Scores: A Systematic Evaluation of Agents for Long-Horizon AI Research and Development

Autonomous agents are increasingly capable of improving models, systems, and other technical artifacts through long-horizon experimentation. To understand the current state of this capability, however, evaluation must go

推荐理由Beyond Final Scores: A Systematic：Autonomous agents are increasingly capable of improving models, systems, and other。
研究与学习HF daily papers
huggingface.co打开原文
#18

SimpleOPD: Simple Tokenizer-Agnostic On-Policy Distillation for Long-Context Reasoning

On-policy distillation (OPD) offers a promising way to transfer reasoning capabilities from stronger teacher models, but applying it to long-context reasoning teachers and short-context students introduces practical chal

推荐理由SimpleOPD: Simple Tokenizer-Agnostic：On-policy distillation (OPD) offers a promising way to transfer reasoning capabilities。
研究与学习HF daily papers
huggingface.co打开原文
#22

DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data

Current large language model development relies on massive, often non-permissible datasets, creating a high barrier for researchers committed to open-source and ethically sourced data. We introduce Mimir v1, a 1-billion

推荐理由DFM Mimir v1: An Open HRM Delivering：Current large language model development relies on massive, often non-permissible。
研究与学习HF daily papers
huggingface.co打开原文
#27

How Do Agents Fail on AutoResearch: End-to-End Diagnostic Evaluation on 100 Real-World Frontier Research Tasks

AI has long assisted scientific research, but the rapid advance of LLMs and agentic scaffolds is reshaping the landscape; a single system can now carry whole-stage research from an initial hypothesis all the way to final

推荐理由How Do Agents Fail on AutoResearch：AI has long assisted scientific research, but the rapid advance of LLMs and agentic。
研究与学习arXiv
arxiv.org打开原文
#43

LLMs Can Predict Failure Risk, But Struggle to Predict Which Collaboration Protocol Pays Off: Cost-Aware Protocol Routing Across Reasoning T

Multi-agent large language model (LLM) systems can improve reasoning by spending more computation, but deployment requires deciding when extra collaboration is worth its cost. We isolate this decision by running every pr

推荐理由LLMs Can Predict Failure Risk, But：Multi-agent large language model (LLM) systems can improve reasoning by spending more。
研究与学习arXiv
arxiv.org打开原文
#46

Skill Blocks: How Should an Agent Load Its Skill? A Caching-Correct Comparison of Pre-load, On-Demand Tool-Loading, Progressive Disclosure,

Agent skills are often injected in full on every request, increasing token cost. We compare four content-preserving loading methods: Full, Skill Block, Reference, and Hybrid. Across SearchQA, SpreadsheetBench, ALFWorld,

推荐理由Skill Blocks: How Should an Agent Load：Agent skills are often injected in full on every request, increasing token cost。
研究与学习arXiv
arxiv.org打开原文
#55

SKILLER: Language-Level Reinforcement Learning for Reusable Skill Extraction in Small Language Models

Agent skills represent a standardized format for packaging procedural knowledge and domain expertise, serving within agent harness systems as an essential mechanism to continually constrain a language model's behavior sp

推荐理由SKILLER: Language-Level Reinforcement：Agent skills represent a standardized format for packaging procedural knowledge。
研究与学习HF daily papers
huggingface.co打开原文
#78

How Can Rhetoric Reward-Hack AI Reviewers? Dissecting Rhetorical Sensitivity in AI-Based Peer Review

As large language models increasingly participate in scientific evaluation, we investigate a potential form of reward hacking: how rhetorical choices shape AI-review judgments when reported scientific content is preserve

推荐理由How Can Rhetoric Reward-Hack AI：As large language models increasingly participate in scientific evaluation, we。
研究与学习HF daily papers
huggingface.co打开原文
#79

PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives

Video world models simulate future states conditioned on current observations and user actions. Recent systems have demonstrated impressive video consistency and action controllability over long sequences. However, fairl

推荐理由PlayWorld: Benchmarking World Models：Video world models simulate future states conditioned on current observations and user。
研究与学习HF daily papers
huggingface.co打开原文
#80

H2R-Bench: Benchmarking Human-to-Robot Manipulation Video Generation in World Models

Large-scale manipulation data is essential for robot learning, yet collecting robot demonstrations remains expensive and difficult to scale. Meanwhile, abundant egocentric human manipulation videos provide rich behaviora

推荐理由H2R-Bench: Benchmarking Human-to-Robot：Large-scale manipulation data is essential for robot learning, yet collecting robot。
研究与学习HF daily papers
huggingface.co打开原文