2026-07-26

AREX: Towards a Recursively Self-Improving Agent for Deep Research

👍 127

07/23 08:00

Deep research requires agents to find answers that jointly satisfy multiple constraints. Discovering such answers is costly, whereas verifying a candidate can often be decomposed into tractable constraint-wise checks. This discovery--verification asymmetry suggests that a research agent should do mo

中文介绍 针对深度研究中“发现-验证不对称”问题（验证易而发现难），提出递归自我改进智能体AREX，通过迭代式探索与验证来高效满足多重约束，显著提升复杂查询的准确率。

SLAI T-Rex: Full-Parameter Post-training of the DeepSeek-V4 Family on Ascend SuperPOD

👍 56

07/22 08:00

Full-parameter post-training of trillion-parameter-scale MoE models introduces substantial system-level challenges for large-scale distributed training, including severe memory pressure, non-overlapped communication overhead, and inefficient kernel execution. While most large-scale LLM training syst

中文介绍 面向万亿参数MoE模型DeepSeek-V4在Ascend SuperPOD上的全参数后训练，提出SLAI T-Rex，通过优化内存、通信和内核执行解决大规模分布式训练挑战，实现高效后训练。

K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs

👍 55

07/23 08:00

Large language models are increasingly used in K-12 education, but existing benchmarks mainly test exam question answering rather than understanding how curriculum knowledge is structured and visually presented. We call this capability curriculum cognition. It covers prerequisite chains, concept tax

中文介绍 提出K12-KGraph，一个课程对齐的知识图谱，用于基准测试和训练K-12教育LLM。它不仅测试考试问答，还评估“课程认知”能力（前提链、概念层次等），为教育模型提供结构化评估。

ReferTrack: Referring Then Tracking for Embodied Visual Tracking

👍 49

07/22 08:00

Embodied visual tracking (EVT) requires a mobile agent to continuously follow a specific target described in natural language using only onboard vision. While recent vision-language-action (VLA) policies unify target identification and trajectory planning, their chain-of-thought (CoT) reasoning ofte

中文介绍 针对具身视觉跟踪中VLA策略的链式推理易出错问题，提出ReferTrack“先指后跟”框架，分离目标定位与轨迹规划，在实机跟踪任务上提升鲁棒性和准确性。

Visual Contrastive Self-Distillation

👍 41

07/23 08:00

On-policy self-distillation (OPSD) is promising as it removes the external teacher required by on-policy distillation (OPD), yet it still needs asymmetric information between teacher and student to ensure that the self-teacher provides a stronger learning signal than the student. Existing methods cr

中文介绍 提出视觉对比自蒸馏（VCSD），在不依赖外部教师的情况下，通过对比框架构建师生间的非对称信息，使自教师提供强于学生的学习信号，在多种视觉任务上有效提升性能。

Show, Don't Tell: Evaluating Spatial Cognition in Generative Pixels Rather Than LLM Text

👍 35

07/23 08:00

Spatial intelligence is essential for agents to move from static semantic understanding toward interacting with the physical world. Many spatial tasks are grounded in continuous visual scenes, where locations, regions, and paths are more naturally expressed by pointing, marking, or drawing than by r

中文介绍 指出空间智能评估应通过生成像素（标记、绘图）而非LLM文本，提出“Show, Don‘t Tell”范式，在连续视觉场景中更自然地评估位置、区域和路径理解，优于传统文本问答。

Beyond Relevance-Centric Retrieval: Rubric-Oriented Document Set Selection and Ranking

👍 28

07/22 08:00

As large language models and AI agents become the primary consumers of search results, document set quality determines the upper bound of downstream generation. Yet existing evaluation systems remain confined to scoring documents independently and aggregating via nDCG, ignoring inter-document intera

中文介绍 超越以相关性为中心的检索，提出基于评分标准的文档集选择与排序方法，考虑文档间交互，避免独立评分聚合带来的局限，提升下游生成质量。

An Exam for Active Observers

👍 24

07/17 08:00

Human vision is a closed loop: gaze is continuously redirected by intermediate hypotheses rather than a single snapshot. Decades of psychophysics and cognitive science have argued that this active observation is essential for a wide range of tasks. Whether today's multimodal large language models (M

中文介绍 为多模态大模型设计“主动观察者”考试，测试其是否具备闭环视觉能力——即根据中间假设动态调整注视点，而非仅处理静态图像。实验发现现有模型在主动观察任务上表现远逊于人类。

NVIDIA-labs OO Agents: Native Python Object-Oriented Agents

👍 22

07/22 08:00

Traditional agent development is split across prompt templates, tool schemas, callback code, and workflow graphs. We present NVIDIA Object-Oriented Agents (NOOA), a model-agnostic Python framework for building reliable AI agents. NOOA takes a simpler approach: an agent is a Python object. Its method

中文介绍 提出NVIDIA Object-Oriented Agents（NOOA），一个模型无关的Python框架，将智能体定义为Python对象，其方法直接映射为工具，统一提示模板、模式、回调和工作流，大幅简化开发。

Tencent WorkBuddy Bench: A Multi-Domain Coding-Agent Benchmark with Contamination-Resistant Task Construction

👍 20

07/23 08:00

We introduce Tencent WorkBuddy Bench, a multi-domain evaluation suite for coding agents; this report documents its construction methodology, scoring protocol, and a cross-model leaderboard. At its core is a unified evaluation framework for constructing and running distribution-informed coding-agent

中文介绍 推出Tencent WorkBuddy Bench，多领域编码智能体评估套件，采用抗污染任务构建方法（分布感知架构），提供标准化评分协议和跨模型排行榜，为编码智能体提供可靠基准。

Color Pass-Through via Camera-Display Coupling

👍 18

07/14 08:00

When a real-world scene is captured by a smartphone camera and viewed on its screen, the displayed image often differs noticeably from the original scene in color, brightness, and contrast. This gap persists despite substantial advances in both modern cameras and displays. A key reason is that most

中文介绍 针对手机拍摄与屏幕显示间的色彩偏差，提出相机-显示器耦合的“颜色通透”方法，通过联合优化捕获和显示环节，实现更忠实的色彩、亮度和对比度还原。

LLMs Get Lost in Evolving User Intent

👍 18

07/22 08:00

As LLMs become more capable, they are increasingly deployed as collaborative agents, taking on user-delegated tasks through iterative interaction. Yet genuine interaction is inherently dynamic: users rarely specify their intent upfront, instead disclosing, revising, and reshaping it as the conversat

中文介绍 研究发现LLM在迭代交互中难以跟踪动态演变的用户意图：用户很少预先完整说明意图，而是不断披露和修正，导致LLM在长对话中迷失方向，降低任务完成质量。

Self-Supervised Learning of Structured Dynamics from Videos

👍 16

07/23 08:00

Understanding motion in video is a fundamental challenge for visual learning, as frame-to-frame change entangles two sources of dynamics: camera motion and object motion. This decomposition has remained underexplored in representation learning, partly because these factors are tightly coupled in nat

中文介绍 从视频中自监督学习结构化动态，将帧间变化分解为相机运动与物体运动两个因素，在无监督条件下学习解耦表示，为视频理解提供更精细的运动建模。

Scaling Laws for Hypernetwork-Based Knowledge Injection in Large Language Models

👍 14

07/21 08:00

Injecting factual knowledge into large language models (LLMs) reliably and at scale remains an open challenge. Hypernetworks provide a promising solution to large-scale knowledge injection. Although hypernetworks are typically applied for test-time adaptation, we explore their use in train-time know

中文介绍 研究基于超网络的知识注入缩放定律，在训练时将事实知识注入LLM。发现超网络能有效扩展知识容量，但注入效果受模型规模、知识粒度等因素影响，为可扩展知识更新提供理论指导。

Beyond Euclidean Clipping: Overcoming Exploration Collapse in LLM RL via Riemannian Isometric Policy Optimization

👍 12

07/11 08:00

Reinforcement learning (RL) has become a dominant paradigm for enhancing LLMs' reasoning capabilities. However, RL algorithms with PPO-Clip are inherently limited by exploration collapse. Subsequent works remain primarily heuristic and fail to identify the essential cause of PPO-Clip's failure. This

中文介绍 揭示PPO-Clip导致LLM强化学习中探索崩溃的本质原因，提出黎曼等距策略优化（RIPO），通过在黎曼流形上保持策略更新的等距性质，克服探索崩溃，提升推理能力。

Streaming Multi-Agent Autoregressive Diffusion Model with World State Registers

👍 11

07/23 08:00

Multi-agent interactive world models should not only generate consistent observations, but also maintain world states that persist across agents and evolve across views. Existing autoregressive video diffusion pipelines carry forward observation history as conditioning context, which makes shared st

中文介绍 提出流式多智能体自回归扩散模型，引入“世界状态寄存器”以保持跨智能体和跨视角的持久化、演进的世界状态，克服传统视频扩散模型中共享状态丢失的问题。

Sample-Efficient Learning from Agent Experience

👍 9

07/23 08:00

Real-world agent learning is often constrained by costly environment interactions, such as running time-consuming experiments or obtaining human feedback. In-context learning offers a highly sample-efficient way for agents to learn from their own interaction histories, but its gains disappear once t

中文介绍 解决基于上下文学习的智能体从自身经验中学习时，效果随交互序列增长而衰减的问题。提出一种样本高效方法，通过结构化经验重用保持学习信号，显著提升长程任务中的样本效率。

AutoIndex: Learning Representation Programs for Retrieval

👍 9

07/21 08:00

We present AutoIndex, a framework for learning representation programs: executable transformations that map raw documents into the representations exposed to a retrieval system. Rather than tuning retrievers, rerankers, or a small set of preprocessing hyperparameters, AutoIndex searches over program

中文介绍 提出AutoIndex框架，学习可执行的表示程序将原始文档映射为检索系统的输入表示，而非调优检索器或超参数，通过在程序空间中搜索找到最优映射，提升检索精度。

Multi-Turn On-Policy Distillation with Prefix Replay

👍 8

07/16 08:00

We study on-policy distillation (OPD) for agentic tasks, where an LLM agent interacts with an environment over multiple turns and a student imitates a teacher over these multi-turn interaction histories. Fully online OPD is costly because each update requires fresh student rollouts through the envir

中文介绍 针对多轮智能体任务的在线策略蒸馏，提出前缀回放方法：存储教师多轮交互历史的前缀，学生基于前缀回放继续交互，大幅降低在线rollout成本，同时保持蒸馏效果。

Robostral Navigate

👍 8

07/22 08:00

Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems depend on depth sensors, multi-camera rigs, or pre-built maps, limiting the hardware they support and increasing deployme

中文介绍 提出Robostral Navigate，一种无深度传感器、不依赖多相机或预建地图的导航方案，通过高效训练策略在多种机器人平台上泛化，实现低成本大规模部署。

The complete Graph Engineering playbook for Claude Code

@Gyome1_ · 3.9K 粉丝 · 220.0K 阅 · 504 赞 · 65 转

07/23 16:53

Most people are still using Claude Code like a very expensive intern. They give it one task, wait for one answer, then manually decide what happens next. But the teams getting the most leverage from

中文介绍 多数人像雇佣昂贵实习生一样使用Claude Code（单任务、等回复、手动决定），而高效团队采用图工程方法，通过构建任务连接图来最大化杠杆效应。分享完整图工程操作手册。

Frontier Diffusion & Control

@satyanadella · 7.5M 粉丝 · 153.7K 阅 · 923 赞 · 114 转

07/24 00:31

In a world where software has real marginal cost for the first time, how do we ensure frontier benefits are diffused across the entire ecosystem? The key is to optimize the cost-to-outcome frontier in

中文介绍 微软CEO Satya Nadella探讨当软件首次拥有实际边际成本时，如何将前沿技术红利扩散至整个生态系统。核心是优化成本-成果前沿，在控制与开放之间找到平衡。

this system will change your life...

@eptwts · 116.8K 粉丝 · 140.0K 阅 · 514 赞 · 28 转

07/24 01:21

the consensus on self-hosted agents is that they're simply coding tools... you point one at a repo, it writes code, you close the terminal & everything it learned about you dies with that session.

中文介绍 批判自托管AI代理的局限：它们只是编码工具，会话结束后对你的了解全部丢失。提出一种能持续学习、记忆用户偏好的系统，打破「用完即弃」模式，实现持久化智能。

Trajectory: A Standard Format for Agent Experience Data

@Letta_AI · 9.8K 粉丝 · 86.5K 阅 · 504 赞 · 31 转

07/24 14:07

Introducing trajectory, an open-source package that normalizes coding-agent sessions from @AnthropicAI Claude Code, @OpenAI Codex, @pidotdev, @LangChain deepagents, @openclaw, Letta Code, and other

中文介绍 Letta AI发布开源包Trajectory，用于标准化来自Claude Code、Codex、deepagents等多个编码代理的会话数据格式，旨在统一代理经验数据，便于分析、复现和跨平台迁移。

Why Software Factories Fail

@dexhorthy · 23.8K 粉丝 · 64.9K 阅 · 515 赞 · 50 转

07/25 00:51

or: the harness is not enough Update - the talk version of this post is live on youtube: https://www.youtube.com/watch?v=Ib5GBkD555M i guess we doin loops now We're all racing to put AI coding into production. A lot has

中文介绍 分析“软件工厂”模式失败的原因：仅靠工具（harness）不够。当前企业急于将AI编码投入生产，但忽略了流程与认知协作，强调需要像循环迭代而非线性流水线来部署AI编码。

The new rules of context engineering for Claude 5 models

@trq212 · 318.0K 粉丝 · 48.6K 阅 · 1.2K 赞 · 119 转

07/25 01:45

I’ve written previously about how to best prompt the newest generation of Claude 5 models and work with them iteratively to discover what you want to build. But when you send a message to Claude, the

中文介绍 面向Claude 5模型的新一代上下文工程规则。区别于传统prompt工程，强调如何结构化上下文、迭代探索需求，通过上下文设计引导模型理解复杂意图，而非单次指令。

Practical multi-agent orchestration in Codex

@pvncher · 30.4K 粉丝 · 44.0K 阅 · 565 赞 · 33 转

07/25 01:30

GPT-5.6 Sol gets especially interesting when it has a team to work with. Codex's new Multi-Agent V2 tools give Sol and Terra a natural way to delegate tasks, share updates, and coordinate through

中文介绍 介绍Codex中实用的多智能体编排能力。Multi-Agent V2工具允许Sol和Terra等代理通过自然方式委派任务、分享更新和协调工作，展示多代理协作的实际操作与价值。

I Can Finally Talk to ChatGPT the Way I Actually Think

@derrickcchoi · 9.1K 粉丝 · 40.1K 阅 · 516 赞 · 31 转

07/24 04:08

This must be a little like how Tony Stark feels talking to Jarvis. I'm really excited that we just launched ChatGPT Voice across Work and Codex in the ChatGPT desktop app. I don't know about you but

中文介绍 ChatGPT桌面版上线Work和Codex语音功能，用户可像与智能助手对话一样自然交流。作者类比托尼·斯塔克与贾维斯，强调语音交互让思考与表达更无缝、更自然。

How we teach AI models

@leerob · 272.8K 粉丝 · 32.9K 阅 · 550 赞 · 51 转

07/24 09:39

How do AI models learn new skills and behaviors? The process is surprisingly human and easy to understand, even if you don’t have a machine learning background. Helpful colleagues AI models,

中文介绍 通俗解释AI模型学习新技能和行为的过程，类比人类与同事协作。无需机器学习背景，核心是通过反馈、示例和迭代训练模型，强调“教学”而非纯编程。

Building the Next Layer of the Agent Economy on BNB Chain

@OptimaiNetwork · 99.3K 粉丝 · 6.1K 阅 · 509 赞 · 112 转

07/23 23:10

Over the past year, AI has moved beyond chat interfaces toward autonomous systems capable of reasoning, planning, and taking action. The next challenge is no longer building more capable agents - it

中文介绍 探讨在BNB Chain上构建代理经济的下一层。AI已从聊天界面进化到自主推理规划系统，当前挑战不再是构建更强大的代理，而是为其创建经济基础设施（如支付、任务分配）。

[AINews] Claude Opus 5: Fable-level performance at Opus price (half Fable)

Latent Space · 07/25 15:25

ain't nobody beats Anthropic at distilling Fable!

中文介绍 Anthropic推出Claude Opus 5模型，在Opus价格下达到Fable级性能（价格仅为Fable的一半）。该模型在基准测试中表现突出，尤其在编码代理能力上获得好评。

Opus 5

Smol AI News · 07/24 13:44

**Anthropic** launched the **Claude Opus 5** model, which sparked mixed reactions including benchmark scrutiny and praise for its coding-agent capabilities. The model achieved an **Epoch Capabilities Index (ECI) of 159**, slightly below **Fable 5's 161**, but matched Fable 5 on software engineering

中文介绍 Anthropic发布Claude Opus 5模型，引发混合反响：基准测试受到质疑，但编码代理能力获赞。模型ECI得分为159，略低于Fable 5的161，但在某些方面与Fable 5持平。

[AINews] Black Forest Labs FLUX 3 - Multimodal Flow Models that beat Seedance 2.0, Gemini Omni and Grok Imagine, and FLUX-mimic video-action robotics model

Latent Space · 07/24 12:30

A HUGE win for BFL!

中文介绍 Black Forest Labs发布FLUX 3多模态流模型，性能超越Seedance 2.0、Gemini Omni和Grok Imagine。同时推出FLUX-mimic视频动作机器人模型，被视为BFL的重大胜利。

ChatGPT Health 🩺, Fugu-Ultra 1.1 🐡, Runway Media Router 🎬

TLDR AI · 07/24 08:00

中文介绍 当日AI新闻简报涵盖三大主题：ChatGPT Health功能更新、Fugu-Ultra 1.1版本发布、以及Runway Media Router的推出。

How AI helps scientists design the next generation of medicines

MIT Tech Review AI · 07/23 20:00

Designing and developing a new medicine is an expensive, failure-prone scientific challenge. A new drug can take many years to develop, at the cost of a significant investment. And even then, most possible candidates never reach the patient. For biologic medicines, therapies made from engineered pro

中文介绍 文章探讨AI如何帮助科学家设计新药。新药研发昂贵且易失败，耗时多年且投入巨大，多数候选药物最终无法惠及患者。AI有望加速生物药等复杂疗法的设计过程。

not much happened today

Smol AI News · 07/23 13:44

**The Stack v3** is released as the largest open code dataset with **114 TB raw data**, **224M repositories**, and **5T deduplicated tokens**, significantly expanding data for open code models and cyber-defense. The debate on **distillation** continues as a key ideological fault line, with calls for

中文介绍 The Stack v3发布，成为最大开源代码数据集，包含114TB原始数据、2.24亿个仓库、5T去重token，极大扩展了开放代码模型和网络防御的数据基础。关于知识蒸馏的争论持续。

[AINews] "Laguna S 2.1 Released: Cheaper than Deepseek v4 Flash, Better than V4 Pro"

Latent Space · 07/23 13:18

a quiet day lets us highlight a new neolab win.

中文介绍 Laguna S 2.1版本发布，价格低于Deepseek v4 Flash，性能优于V4 Pro。这是Neolab的新成果，在平静的一天中成为亮点。

Inside the Model Factory — Eiso Kant, Poolside AI

Latent Space · 07/23 13:09

Poolside's co-CEO on how his small team of top researchers built a model factory capable of training Laguna S - a 118B MOE beating Thinky's ~1T open weights model... and this is just the beginning.

中文介绍 Poolside AI联合CEO Eiso Kant介绍其小团队如何建立模型工厂，训练出118B参数的MOE模型Laguna S，该模型性能击败了Thinky约1T参数的开源模型。这仅是开始。

Launching Health in ChatGPT

OpenAI News · 07/23 08:00

Health in ChatGPT now lets eligible U.S. users securely connect medical records and Apple Health to get more personalized insights and better understand their health.

中文介绍 OpenAI在ChatGPT中推出Health功能，允许符合条件的美国用户安全连接医疗记录和Apple Health数据，以获得个性化健康洞察。

Bringing Nunchaku 4-bit Diffusion Inference to Diffusers

Hugging Face Blog · 07/23 08:00

中文介绍 Hugging Face博客介绍将Nunchaku 4位扩散推理引入Diffusers框架，旨在提高图像生成模型的推理效率。

Cursor Router 🔀, OpenAI Presence 🤖, AMD + Anthropic deal 🤝🏻

TLDR AI · 07/23 08:00

中文介绍 当日AI简报涵盖Cursor Router发布、OpenAI Presence功能更新、以及AMD与Anthropic达成合作协议。

Accelerating the frontiers of scientific discovery: Google’s $40M commitment to the Genesis Mission

DeepMind Blog · 07/22 21:38

Google commits $40M in AI tokens and credits for the Genesis Mission

中文介绍 谷歌承诺为Genesis任务提供4000万美元的AI token和积分，以加速科学发现的前沿探索。

Building AI infrastructure with the Effingham County community

OpenAI News · 07/22 21:00

OpenAI announces Project Camellia in Effingham County, Georgia, with commitments to responsible energy, community investment, jobs, and access to Codex.

中文介绍 OpenAI宣布在美国佐治亚州Effingham县启动Project Camellia项目，承诺负责任能源、社区投资、就业以及提供Codex访问权限。

How news organizations are using AI to advance their vital missions

OpenAI News · 07/22 21:00

News organizations are using AI to strengthen reporting, grow audiences, and improve business operations, with OpenAI tools supporting journalists and publishers worldwide.

中文介绍 新闻机构利用AI加强报道、扩大受众、改善运营。OpenAI工具正在全球范围内支持记者和出版商实现其重要使命。

Advancing the next era of national science

OpenAI News · 07/22 20:00

OpenAI outlines its commitment to advancing American science working with the U.S. Department of Energy and national labs to use frontier AI to accelerate discovery.

中文介绍 OpenAI阐述其与美国能源部及国家实验室合作，利用前沿AI加速科学发现，推动美国科学进入新时代的承诺。