2026-07-15

Weak-to-Strong Generalization via Direct On-Policy Distillation

👍 93

07/08 08:00

Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training. As models scale, post-training itself becomes a bottleneck.

中文介绍 针对RLVR重复训练新模型成本高的问题，提出直接在线策略蒸馏（Direct On-Policy Distillation），通过弱模型引导强模型实现弱到强泛化，显著降低后训练开销。

ABot-N1: Toward a General Visual Language Navigation Foundation Model

👍 81

07/12 00:21

Visual Language Navigation foundation models aim to unify deep reasoning for grounded spatial decisions with broad versatility for diverse embodied tasks. Current approaches typically achieve this integration via monolithic policies that map observations directly to actions, yet they often suffer fr

中文介绍 提出ABot-N1通用视觉语言导航基础模型，统一空间决策与深度推理，克服现有单策略方法泛化不足的问题，在多项具身导航任务上表现出色。

ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory

👍 68

07/11 23:24

Recent VLM and VLA systems have improved robotic perception and action prediction, yet long-horizon embodied agents still require a general runtime layer for reasoning, memory, tool use, verification, and cross-embodiment execution. We present ABot-AgentOS, a general robotic Agent Operating System t

中文介绍 提出ABot-AgentOS通用机器人智能体操作系统，提供终身多模态记忆与推理、工具调用、验证等运行时层，支持跨实体长期任务执行，弥补VLM/VLA在长时自主决策上的不足。

Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

👍 61

07/09 08:00

AI agents have become capable of autonomously completing short, well-specified tasks. However, existing terminal benchmarks largely focus on simple problems that finish within minutes and are evaluated only by their final outcome. This setup overlooks intermediate progress and partial solutions, yie

中文介绍 提出Long-Horizon-Terminal-Bench基准，针对长期终端任务采用密集奖励评分体系，评估智能体在数小时级任务中的中间进展与最终表现，弥补现有基准仅关注短期结果的不足。

Video-Oasis: Rethinking Evaluation of Video Understanding

👍 61

07/02 08:00

The inherent complexity of video understanding makes it difficult to determine whether Video-LLM benchmark performance stems from visual perception, linguistic reasoning, or knowledge priors. While many benchmarks have emerged to assess high-level reasoning, shared criteria for evaluating video unde

中文介绍 针对视频理解基准难以区分视觉感知、语言推理与知识先验的问题，提出Video-Oasis评估框架，通过去耦设计分离各能力贡献，为视频LLM提供更细粒度的评估标准。

Video Generation Models are General-Purpose Vision Learners

👍 54

07/10 08:00

Driven by next-token prediction, NLP shifted from task-specific models into powerful generalist foundation models. What, then, is the equivalent catalyst needed to achieve a general-purpose model in computer vision? In this paper, we contend that large-scale text-to-video generation serves as a stro

中文介绍 类比NLP中下一词预测催生通用模型，论证大规模文本到视频生成可作为通用视觉学习范式，预训练的视频生成模型在多种下游视觉任务上展示强迁移能力，迈向通用视觉基础模型。

Scalable Visual Pretraining for Language Intelligence

👍 49

07/10 08:00

The rapid progress of large foundation models has been driven predominantly by pretraining on large-scale text corpora. However, many forms of knowledge are conveyed through visual representations, where figures, typeset equations, and page layouts carry rich information that cannot be faithfully or

中文介绍 针对文本预训练难以捕获图表、公式等视觉知识的问题，提出可扩展的视觉预训练方法，从页面布局、图示中学习结构化信息，增强语言模型在需要视觉理解的推理任务上的表现。

UniClawBench: A Universal Benchmark for Proactive Agents on Real-World Tasks

👍 32

07/09 08:00

The rapid development of large language models and multimodal large language models has accelerated the emergence of proactive agents capable of operating everyday tools and assisting users in real-world environments. However, existing benchmarks struggle to evaluate such agents effectively, as they

中文介绍 提出UniClawBench通用基准，用于评估主动智能体在真实世界操作工具和辅助用户的能力，覆盖多种日常任务场景，弥补现有基准在环境交互与工具使用方面的评估缺失。

LongE2V: Long-Horizon Event-based Video Reconstruction, Prediction, and Frame Interpolation with Video Diffusion Models

👍 31

07/09 08:00

Recovering high-quality video from sparse event streams is a challenging task. Regression methods often blur textures, while existing generative models struggle with long-term stability. We propose LongE2V, a novel approach that leverages pre-trained video diffusion priors to jointly handle event-ba

中文介绍 针对事件相机稀疏流重建视频质量差、长期不稳定问题，提出LongE2V，利用预训练视频扩散模型先验，联合处理事件到视频重建、预测和帧插值，在长时域上保持纹理清晰与时间一致性。

Trust Region Policy Distillation

👍 22

07/06 08:00

Big goals are hard to achieve all at once; breaking them into small steps is wiser. We present Trust Region Policy Distillation (TOP-D), which transforms the notoriously unstable, high-variance On-Policy Distillation (OPD) into a stable training paradigm by dynamically constructing a proximal teache

中文介绍 提出信任域策略蒸馏（TOP-D），将高方差、不稳定的在线策略蒸馏转化为稳定训练范式，通过动态构建近端教师保证更新步长在信任域内，显著提升LLM强化学习后训练的样本效率与收敛稳定性。

AdvancedMathBench: A Benchmark Suite for Advanced Mathematical Proof Generation and Verification

👍 20

07/14 01:38

Large language models (LLMs) have achieved remarkable performance on high-school and olympiad-style mathematics, yet their capabilities on advanced mathematics remain poorly understood. Existing benchmarks, however, fall short in both scope and evaluation granularity: they provide limited disciplina

中文介绍 针对LLM在高等数学（如数论、拓扑）证明能力评估不足，提出AdvancedMathBench基准套件，包含多学科高级数学问题，支持证明生成与验证的细粒度评测，揭示模型在形式推理上的局限。

Enhancing In-context Panoramic Generation via Geometric-aware Pretraining

👍 20

07/09 08:00

In this work, we present Canvas360, a two-stage framework for in-context panoramic generation that combines geometry-aware pretraining with downstream task-specific fine-tuning. To address the lack of large-scale, high-quality training data tailored to in-context panoramic tasks, we propose Canvas36

中文介绍 针对上下文全景生成缺乏高质量训练数据，提出Canvas360两阶段框架：先进行几何感知预训练学习空间结构，再微调适配下游任务；配合数据生成管道，在多项全景生成任务上提升一致性。

KronQ: LLM Quantization via Kronecker-Factored Hessian

👍 19

07/08 08:00

Post-training quantization (PTQ) is a widely adopted technique for compressing large language models (LLMs) without retraining. Existing second-order PTQ methods, including GPTQ, construct quantization objectives exclusively from input activation statistics, effectively assuming that all output chan

中文介绍 针对后训练量化（PTQ）忽视输出通道差异的问题，提出KronQ，通过克罗内克分解近似Hessian矩阵，构建更精确的量化目标，在LLM上实现更低精度损失，优于GPTQ等现有二阶方法。

CineMobile: On-Device Image-to-Video Diffusion for Cinematic Camera Motion Generation

👍 15

07/04 08:00

The growing demand for image-to-video creation on mobile devices has increasingly focused on cinematic motion effects like bullet time, dolly zoom, slow motion, etc. While Diffusion Transformers (DiTs) exhibit strong performance in video generation, their large parameter sizes and multi-step iterati

中文介绍 面向移动设备上电影级镜头运动（如推拉、慢动作）生成需求，提出CineMobile，轻量化图像到视频扩散模型，通过模型压缩与单步采样实现设备端实时生成，保持运动质量。

Metacognition in LLMs: Foundations, Progress, and Opportunities

👍 12

07/14 01:58

Metacognition is a foundational component of intelligence critical to effective learning, problem solving, decision-making, communication, and more. In recent years, it has become increasingly recognized as a cornerstone of capable, transparent AI systems. Yet while LLMs have made significant progre

中文介绍 系统综述大语言模型中的元认知能力，涵盖自我监控、反思、策略调整等维度，总结现有进展（如链式思考、自我评估）及评估方法，指出当前局限与未来研究方向。

Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuning

👍 11

07/09 20:17

Fine-tuning LLMs to inject new knowledge faces a critical challenge: LLMs can quickly memorize new facts, yet fail to use them for downstream reasoning tasks. We formalize this failure as the \textbf{Knowing--Using Gap}, characterized by an accuracy gap and a temporal lag between memorization and ge

中文介绍 定义微调中LLM“知道-使用”差距（记忆但不泛化），通过机制分析揭示知识存储在特定参数通路中，但与推理路径耦合不足；提出缓解策略，弥合记忆到泛化的鸿沟。

Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents

👍 10

07/09 08:00

In long-horizon tasks, decision-relevant state is often scattered across an expanding trajectory, while the action agent must surface it and act. As trajectories grow, task requirements, environment facts, prior attempts, diagnoses, and open subgoals can be buried in the context window or pushed bey

中文介绍 针对长期任务中智能体遗忘关键状态（如环境事实、子目标）的问题，提出主动记忆智能体，实时识别并注入相关记忆，在长轨迹中维持任务上下文，显著提升长时自主决策成功率。

Self-Guided Test-Time Training for Long-Context LLMs

👍 9

07/10 08:00

Long-context processing has become increasingly important for large language models (LLMs), but simply extending the context window does not guarantee effective utilization of long inputs. As input length grows, accuracy often degrades, indicating that models still struggle to identify and use the e

中文介绍 针对LLM虽支持长上下文但实际利用率低的问题，提出自引导测试时训练，在推理阶段动态识别输入中的关键位置并调整注意力，显著提升长文档理解与信息检索的准确率。

UP: Unbounded Positive Asymmetric Optimization for Breaking the Exploration-Stability Dilemma

👍 8

07/08 08:00

Reinforcement learning (RL) has become the standard paradigm for enhancing the complex reasoning capabilities of large language models (LLMs). To achieve sample efficiency, modern RL frameworks rely on importance sampling (IS). However, these algorithms suffer from an exploration-stability dilemma.

中文介绍 针对RL中重要性采样算法面临的探索与稳定性矛盾，提出UP（无界正非对称优化），通过非对称损失设计解耦策略探索与梯度稳定性，在LLM强化学习后训练中实现更高效的样本利用与更稳定收敛。

Proxy Exploration and Reusable Guidance: A Modular LLM Post-Training Paradigm via Proxy-Guided Update Signals

👍 7

07/13 20:56

Post-training is essential for refining the domain-specific capabilities of large language models (LLMs), yet existing reward optimization and distribution matching methods tightly couple policy exploration with distribution alignment. This coupling forces expensive exploration directly on the polic

中文介绍 针对LLM后训练中策略探索与分布对齐紧密耦合导致高成本的问题，提出模块化范式Proxy Exploration and Reusable Guidance，利用代理模型引导探索生成可重用更新信号，解耦探索与对齐，提升后训练效率与灵活性。

Making Fable Cheaper Than Opus

@joon_h_lee · 515 粉丝 · 366.3K 阅 · 521 赞 · 43 转

07/14 01:03

We replaced Opus 4.8 with Fable 5, and Devin’s bill went down. Fable 5 costs twice as much per token as Opus 4.8. But when we ran both models on FrontierCode 1.1 using our new Fusion architecture,

中文介绍 用Fusion架构替换模型：Fable 5每token成本是Opus 4.8两倍，但在FrontierCode 1.1上运行后Devin账单反而下降。分享模型替代与成本优化实践。

The Most Human Technology Ever Made

@illscience · 30.4K 粉丝 · 332.5K 阅 · 539 赞 · 69 转

07/14 01:05

Over the weekend, Mira Murati wrote a phenomenal essay that you should read in full: “The future worth building is human.” I think it’s one of the best essays of the year, because it gets at the

中文介绍 推荐Mira Murati的essay《The future worth building is human》，称其为年度最佳文章之一，探讨技术应回归人性，是AI与人类未来最深刻的思考。

The Reverse Information Paradox

@satyanadella · 7.0M 粉丝 · 146.5K 阅 · 1.1K 赞 · 162 转

07/12 23:09

In the age of intelligence, how should firms protect their core IP? Nobel Prize winning economist Kenneth Arrow famously described a paradox in the market for information. “Its value for the

中文介绍 Satya Nadella探讨智能时代企业如何保护核心IP，引用肯尼斯·阿罗信息悖论，指出公司需重新思考数据策略以应对价值评估难题。

A Framework for Frontier AI and the Dawning of a New Age

@demishassabis · 1.4M 粉丝 · 98.1K 阅 · 1.5K 赞 · 236 转

07/14 17:10

This is a pivotal moment in human history. Artificial General Intelligence (AGI), a system that exhibits all the cognitive capabilities the brain has, is probably only a few short years away. When we

中文介绍 DeepMind创始人Demis Hassabis称AGI仅数年之遥，提出前沿AI框架，强调人类历史转折点需为新时代做好准备并探索负责任发展路径。

I Brutally Modified My Front-End Design Skill ~ Now My UIs Don’t Look Like AI Crap

@JoePro · 949 粉丝 · 66.8K 阅 · 537 赞 · 29 转

07/14 11:51

To create a custom Claude skill, navigate to Customize > Skills in your Claude interface, click the + button, and select + Create skill

中文介绍 通过自定义Claude Skill提升前端设计质量：进入Customize > Skills创建新技能，让UI摆脱「AI味」，回归专业设计水准。

How I'd make $10 million with AI agents

@gregisenberg · 682.7K 粉丝 · 62.2K 阅 · 578 赞 · 36 转

07/14 02:22

2 markets worth trillions are opening at the same time. I've been through 3 platform shifts living in Silicon Valley building/selling VC-backed companies and I've NEVER seen this. Mobile apps came

中文介绍 硅谷连续创业者Greg Isenberg称两个万亿级市场同时开启，用AI Agent可赚1000万美元，认为当前机遇远超移动App时代。

2 Hermes Workflows I can't live without

@0xJeff · 80.8K 粉丝 · 61.6K 阅 · 502 赞 · 36 转

07/13 19:33

I’ve been consistently using Hermes everyday for the past 4 months and he’s already a core part of my research & investment workflows. Before Hermes, it was just me relying on Grok, Claude, Gemini for

中文介绍 连续4个月使用Hermes后分享两个不可替代工作流：研究和投资。之前依赖Grok、Claude等，Hermes已成核心效率工具。

Motion-Graphics Skill

@HeyGen · 106.3K 粉丝 · 58.5K 阅 · 516 赞 · 56 转

07/13 03:00

Day 7 of 30: Motion Graphics by Example /motion-graphics makes short, design-led motion graphics: usually under 10 seconds, no narration, motion is the message. Kinetic type, stat count-ups, charts,

中文介绍 HeyGen发布Motion Graphics技能：生成10秒内无旁白的动效，支持动态文字、数据计数、图表等，适用于短视频营销与演示。

How To Actually Design With AI

@LexnLin · 17.6K 粉丝 · 32.3K 阅 · 503 赞 · 27 转

07/13 05:44

Design has always been one of the areas where AI struggles to match human ability. But most people still misunderstand what designing with AI actually means. Designing with AI does not mean asking it

中文介绍 LexnLin指出多数人误解AI设计：不是直接出图，而是作为协作工具。AI能力仍难媲美人类，正确用法是辅助而非替代。

Your AI says it read your files. Did it really?

@conikeec · 7.5K 粉丝 · 1.1K 阅 · 1.0K 赞 · 121 转

07/14 22:39

AI always answers with confidence — even when it skipped half your documents. waggle shows you which files it really opened, so you know whether to trust the answer, or send it back. You hand an AI

中文介绍 工具waggle解决AI「假装读文件」问题：显示实际打开的文件列表，用户可判断答案可靠性，适用于文档分析场景。

[AINews] not much happened today

Latent Space · 07/15 07:54

a continuation: Codex adding 1M users a day now.

中文介绍 Codex目前每天新增100万用户。

5 Trends That Defined AI Engineering at World’s Fair 2026

Latent Space · 07/15 07:21

At this year's AIE World’s Fair, AI engineering entered a new phase: building systems around agents, rather than just building with agents.

中文介绍 2026年AI工程世界博览会展示五大趋势，AI工程进入以智能体为中心构建系统的新阶段。

How to manage AI investments in the agentic era

OpenAI News · 07/14 18:00

Learn how enterprises can manage AI investments in the agentic era by measuring useful work per dollar, improving efficiency, and scaling high-value workflows.

中文介绍 企业可通过衡量每美元的有用工作、提高效率及扩展高价值工作流来管理智能体时代的AI投资。

not much happened today

Smol AI News · 07/14 13:44

**OpenAI's agent products** saw a **2.5x weekly usage growth** driven by **Codex + ChatGPT Work** and demand for **GPT-5.6 Sol**. JetBrains adopted Codex as a recommended agent, while LangChain enhanced tracing and observability across multiple tools. **PrismML released Bonsai 27B**, a compressed va

中文介绍 OpenAI代理产品周使用量增长2.5倍，由Codex和ChatGPT Work驱动，并对GPT-5.6 Sol有需求。JetBrains采用Codex为推荐代理，LangChain增强追踪和可观测性，PrismML发布Bonsai 27。

[AINews] Codex usage up >10x in 6 months to 7M users, +1M in the past ~day; did Codex overtake Claude Code??

Latent Space · 07/14 09:22

a quiet day lets us fact check some numbers against the sound of silence of Claude Code reporting...

中文介绍 Codex使用量六个月内增长超10倍，用户达700万，过去一天新增约100万；与Claude Code的对比未明确。

How data science teams use ChatGPT Work

OpenAI News · 07/14 08:00

See how data science teams can use ChatGPT Work to build root-cause briefs, impact readouts, KPI memos, scoped analyses, and dashboard specs from real work inputs.

中文介绍 数据科学团队可用ChatGPT Work从实际工作输入生成根因简报、影响报告、KPI备忘录、范围分析和仪表盘规格。

How sales teams use ChatGPT Work

OpenAI News · 07/14 08:00

See how sales teams can use ChatGPT Work to create pipeline briefs, meeting prep packets, forecast reviews, account plans, and stalled-deal diagnoses from real work inputs.

中文介绍 销售团队可用ChatGPT Work创建管道简报、会议准备包、预测审查、客户计划和停滞交易诊断。

xAI uploads codebases 👨‍💻, Prime Intellect verifiers 🧠, Sakana smart bricks 🧱

TLDR AI · 07/14 08:00

中文介绍 xAI上传代码库，Prime Intellect推出验证器，Sakana发布智能砖块。

What Anthropic’s latest AI discovery does—and doesn’t—show

MIT Tech Review AI · 07/14 02:00

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Anthropic—currently the world’s most valuable AI company, with a nearly $1 trillion valuation—has a reputation for publishing strange and heady research. It’s loo

中文介绍 Anthropic（估值近1万亿美元）的最新AI发现及其局限性探讨。

Empowering India’s next generation of innovators with ATL Saathi

DeepMind Blog · 07/13 20:37

Google and AIM launched ATL Saathi, a Gemini-powered AI tool empowering Indian educators in robotics labs.

中文介绍 Google与AIM联合推出ATL Saathi，一个基于Gemini的AI工具，赋能印度教育工作者在机器人实验室中的教学。

not much happened today

Smol AI News · 07/13 13:44

**Prime Intellect** released **verifiers v1**, a redesigned environment stack for **agentic reinforcement learning** and evaluations, improving efficiency by storing rollout traces as **message DAGs** to reduce complexity from **O(n²)** to **O(n)**. This enables practical long-horizon multimodal rol

中文介绍 Prime Intellect发布verifiers v1，重新设计代理强化学习环境栈，通过将rollout迹存储为消息DAG，将复杂度从O(n²)降至O(n)，提升效率。

Claude Code browser 🌍, Cursor general agent 🤖, Claude Fable extension ⏳

TLDR AI · 07/13 08:00

中文介绍 Claude Code推出浏览器功能，Cursor发布通用代理，Claude Fable扩展上线。

[AINews] not much happened today

Latent Space · 07/11 10:53

a quiet day after a week of nonstop model releases

中文介绍 在连续一周的模型发布之后，今日较为平静。

How Deutsche Telekom is rewiring telecommunications with AI

OpenAI News · 07/10 15:00

How Deutsche Telekom is becoming an AI-native telco with OpenAI-transforming customer service, employee workflows, network operations, and the future of voice.

中文介绍 德国电信正借助OpenAI转型为客户服务、员工工作流、网络运营及语音未来，成为AI原生电信公司。

[AINews] OpenAI launches GPT 5.6 Sol/Terra/Luna, Codex becomes ChatGPT superapp

Latent Space · 07/10 14:19

A big day for OpenAI.

中文介绍 OpenAI发布GPT 5.6 Sol/Terra/Luna，Codex成为ChatGPT超级应用。