2026-10-17

LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents

👍 4

10/15 08:00

Computer-use agents increasingly tackle long-horizon tasks, where running a frontier model at every step is expensive and a small open-weight model alone often fails. Weak-to-strong escalation offers a middle ground where a weak agent acts by default and hands control to a strong one only when neede

Explaining the Roles of Experts in Mixture-of-Experts LLMs for Personalization

👍 2

10/15 08:00

Personalization of Large Language Models (LLMs) is increasingly important for user-centric applications, yet prior methods mainly adapt dense LLMs, whose polysemantic neurons, each encoding multiple unrelated concepts, hinder localizing the components behind personalization, and thus understanding a

When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill

👍 1

10/15 08:00

Reusing the attention KV cache across requests is a standard way to cut prefill cost in long-context LLM serving, and production systems increasingly back the per-engine cache with a remote tier shared by every engine. We study such a tier in a disaggregated deployment of GLM-5.3, an FP8 mixture-of-

SUPRA-TEX: 3D Generation with Super-Aligned Texture

👍 7

10/15 08:00

Single image 3D generation aims to dress a given 3D mesh with relightable materials while faithfully preserving the visual details of a reference image. Existing methods either introduce artifacts when fusing inconsistent multi-view predictions or struggle to preserve fine details through implicit i

FDE-Gym: The Last Mile of AI Delivery

👍 229

10/15 08:00

Existing agent benchmarks typically evaluate agents on well-specified tasks defined by issue reports, specifications, or task tickets. Real-world customer delivery is different: forward-deployed engineers (FDEs) must clarify ambiguous requirements, work in live systems, and uncover unstated constrai

Data World: Scaling Data Environments from Scratch

👍 24

10/15 08:00

Data agents use code to process heterogeneous files and derive analytical insights. Scaling their training faces costly environment curation, privacy and intellectual-property restrictions, and difficulty verifying open-ended findings without known generative truth. Our core insight is to specify th

OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling

👍 27

10/15 08:00

Generative visual verifiers play a crucial role in advancing multimodal intelligence. To enhance the robustness of visual verifiers in complex multimodal scenarios involving compositional requirements, we move beyond conventional binary verifiers that produce only True/False discrete judgments and i

CoWorld Reasoner: Agentic Multimodal Reasoning with the Evolving Video World Model-Driven Harness

👍 1

10/15 08:00

As Vision-Language Models (VLMs) advance, they are increasingly expected to perform spatiotemporal reasoning over motion, actions, and view changes beyond what the input images show. Video World Models (VWMs) provide a natural unified engine for visual imagination by generating task-relevant visual

OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use

👍 266

10/15 08:00

Computer use has reached a turning point: the focus is expanding from GUI-only interaction to systems that complete tasks on users' computers through visual and programmatic interfaces. We define a generalist computer-use agent (GCUA) as a system in which the model chooses how to observe, act, and o

MJev-Spatial: System One Spatial Intelligence via Native Attention Matching

👍 140

10/15 08:00

Spatial intelligence lets an agent perceive and act in the physical world, and many of its judgments must be made at interaction rates: which spatial relation holds, how far an object is, or where to place it. Vision-language models (VLMs) produce these answers by free-form text generation, paying f

VLA-Δ: Learning Physical State Transition for Vision-Language-Action Models

👍 53

10/15 08:00

Vision-language-action (VLA) models transfer semantic knowledge acquired through vision-language pretraining to robot control, achieving strong performance across various benchmarks. However, most VLAs directly map observations to demonstrated actions without explicit supervision of intermediate phy

Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation

👍 451

10/15 08:00

Policy-shift distillation reuses the difference between a teacher's log probabilities before and after reinforcement learning (RL) as supervision on student-generated trajectories. However, this endpoint policy shift cannot reveal whether local action preferences evolved consistently during teacher

MMSearch-OPD: Easy Made Hard, Hard Made Learnable for Long-Horizon Multimodal Search Agents

👍 4

10/15 08:00

Multimodal deep search is moving from short queries to long-horizon investigations that continually acquire evidence. However, training such agents remains constrained by limited challenging training questions and insufficient process feedback. We introduce MMSearch-OPD, combining controllable dat

MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering

👍 24

10/15 08:00

Large language models (LLMs) are expanding from assisting scientific research and engineering to autonomously developing and improving AI systems, advancing AI for AI (AI4AI) as a pathway toward recursive self-improvement (RSI). Workflow-driven machine learning engineering (MLE) systems provide a pr

Learning Self-Orchestration in Decentralized Agent Swarms

👍 9

10/15 08:00

The capabilities of an agent swarm depend on both its individual members and how they coordinate. Yet existing multi-agent systems often train agents within fixed orchestration structures, separating the learning of individual capabilities from the organization of collective behavior. We introduce m

Equal CER, Unequal Damage: Severity-Weighted Character Error Rate for OCR

👍 2

10/14 08:00

Character Error Rate (CER), the standard metric for OCR and vision-language models (VLMs), charges a typo as much as a wrong name. We introduce semCER, which keeps CER's alignment and denominator and charges each grapheme edit a severity from an LLM judge, with no word segmentation. Normalized CER k

Learning to Read Out: Unembedding Dynamics in Language Model Pretraining

👍 2

10/14 08:00

When does a language model acquire a capability? Developmental studies usually answer this by tracking when models start producing the correct token. But that prediction passes through the unembedding matrix, or readout, which is itself still being learned during pretraining. We show that a model's

MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments

👍 93

10/14 08:00

Online reinforcement learning (RL) with verifiable rewards has driven rapid progress in mathematics, coding, and tool use, but scaling this success to GUI agents remains challenging due to the high cost of interactive environments and the scarcity of realistic training tasks. We introduce MAI-Gym, a

SERO: Learning from Executable Oracle Continuations for Long-Horizon Agentic Reinforcement Learning

👍 4

10/14 08:00

Long-horizon agentic RL is constrained by the futures visited by the current policy. When rollouts fail to reach useful behaviors, improved credit assignment cannot provide supervision for states and actions that were never encountered. On-policy distillation (OPD) introduces teacher guidance on pol

WorldCast: Exploring Intelligent Visual Generation in World State Prediction

👍 29

10/14 08:00

Humans interpret multimodal observations of the world and forecast how it will evolve. Recent multimodal generative systems have demonstrated similar capabilities: they can integrate multimodal context and synthesize visual outputs that depict changes in the world. To characterize this emerging traj

Copy-Paste Language Models

👍 3

10/13 08:00

Transformers are intuitively considered to be efficient at copying from their inputs. We actually show that the ability of LLMs to copy in context can be improved through explicit modeling. We propose Copy-Paste Language Models (CPLMs), which add a lightweight Copy-Paste head to the conventional lan

Virtual Head Attention

👍 4

10/13 08:00

The design of attention mechanisms in long-context large language models faces an ``impossible triangle'' among KV cache size, training and prefill FLOPs, and model quality. GQA achieves strong quality but incurs high KV cache overhead; alternative approaches such as MLA and MFA can compress the cac

Cowork-Gym: Synthesizing Interactive Work Environments at Scale

👍 5

10/13 08:00

Cowork agents navigate dependencies across tools and files while adapting to changing user requirements and edits to shared workspaces. Scaling such environments requires maintaining consistency among task objectives, executable resources, user behavior, and success criteria. We introduce Cowork-Gym

H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression

👍 3

10/13 08:00

Training-free diffusion-based compression methods repurpose the generative priors of a diffusion or a flow model to compress data without retraining them as task-specific codecs. Recent codebook-based approaches such as Turbo-DDCM achieve strong perceptual quality at ultra-low bitrates, but their re

Beyond Policy Rollouts: Embedding-Space Directional Exploration with Anchored Noise for LLM RLVR

👍 3

10/12 08:00

Reinforcement learning with verifiable rewards (RLVR) typically uses small rollout groups per prompt, where repeated sampling from the same policy can yield highly similar trajectories, reducing within-group reward variation and weakening group-relative advantage signals. Existing remedies such as h

PINT: A Unified Model with Shared Predictive Intent for Robot Manipulation

👍 2

10/12 08:00

Generalist robot manipulation can benefit from pretrained vision-language models (VLMs) for semantic reasoning and video generation models (VGMs) for world modeling. However, existing unified models that directly fuse them through shared attention at multiple layers can make it harder to draw on the

Recursive Environment Scaling for General Long-Horizon Agents

👍 2

10/12 08:00

We study environment scaling as a way to increase agent-benchmark difficulty while holding core task objectives fixed. Our framework, Recursive Environment Scaling (RES), transforms existing benchmark environments while preserving their objectives and completion requirements. A proposer changes info

Scalable Weak-to-Strong Verifier Hardening for Agent Environments

👍 1

10/11 08:00

Agent environments are essential for large language model (LLM) development. They use programmatic verifiers to turn complex outcomes into rewards. Verifiers can be attacked into awarding full credit to cheating agents without completing the intended task, causing inaccurate evaluation and misalignm

Speech event detector: Timestamped detection of non-verbal vocalizations and background sounds for rich speech transcription

👍 2

10/09 08:00

Automatic speech recognition produces words but drops the non-verbal content of a recording: the breaths, coughs, sighs and laughter of the speaker, and the horns, music or barking around them. We describe Speech Event Detector, a 5.8-million-parameter convolutional-Transformer model that outputs pr

PAIR: Bridging Perception and Action in Vision-Language-Action Models

👍 1

10/06 08:00

Vision-language-action (VLA) models map visual observations and language instructions to continuous robot actions. This task requires a transition from representations that describe the scene and instruction to representations that support action generation. Many continuous-action VLAs leave this tr

Is CoreWeave Going to Die

@goodalexander · 174.8K 粉丝 · 45.4K 阅 · 510 赞 · 27 转

10/16 01:11

TLDR: no. Coreweave I do not think is going to die. I think Anthropic roadshow will do better than expects due to Fable 5.5 lifting sentiment on Lab funding, with possibility of government

中文介绍 分析 CoreWeave 前景，认为其不会倒闭，并看好 Anthropic 的未来发展。

Build or surrender

@arthurmensch · 88.8K 粉丝 · 17.9K 阅 · 619 赞 · 109 转

10/16 15:05

An idea has gained ground in Europe: we should leave frontier AI models to the Americans and the Chinese. It is advocated by serious people, whose reasoning has its logic. A frontier model costs

中文介绍 探讨欧洲关于将前沿 AI 模型留给美中两国的观点，分析其逻辑与成本问题。

What I brought home from OpenAI DevDay

@BradGroux · 6.9K 粉丝 · 1.3K 阅 · 512 赞 · 3 转

10/17 00:23

On the drive home from the airport after OpenAI DevDay, I spent about 45 minutes on a phone call with AVA, my dot. Before I got home, I had planned the next day and moved work forward across several

中文介绍 分享 OpenAI DevDay 的收获，包括与 AVA 的电话沟通和次日工作计划。

Roundtables: A Conversation With the Creator of AI-Designed Viruses

MIT Tech Review AI · 10/17 02:00

Listen to the session or watch below Can AI design new life forms? In 2025, Stanford University PhD student Samuel King came up with a preliminary answer when he used a generative AI model to propose genetic blueprints for microscopic viruses. It isn’t yet an example of AI-generated life, but that c

中文介绍 斯坦福大学博士生Samuel King利用生成式AI模型设计了微观病毒的基因蓝图，探讨了AI设计新生命形式的可能。

[AINews] Agents disprove Polysemanticity; greatly improve Interpretability & Alignment

Latent Space · 10/16 12:16

Nick Cammarata out of nowhere with some major claims!

中文介绍 Nick Cammarata提出重要观点，否定多义性，显著提升可解释性和对齐。

Scaling Indic Pretraining: 2.8 Trillion Tokens and an Open Data Generation Pipeline

Hugging Face Blog · 10/16 09:44

中文介绍 Hugging Face博客发布关于扩展indic预训练的博文，涉及2.8万亿token和开放数据生成管道。

Stack Overflow for Agents triples signups with ChatGPT plugin

OpenAI News · 10/16 08:00

Two engineers shipped Stack Overflow for Agents in a few weeks with Codex, and its ChatGPT plugin helped drive up to 10× growth in agent creation.

中文介绍 OpenAI宣布Stack Overflow for Agents使用ChatGPT插件，数周内用户增长10倍。

Welcome Beam: a 501B open model you can own, customize, and scale

Hugging Face Blog · 10/16 08:00

中文介绍 Hugging Face推出Beam模型，一个可拥有、定制和扩展的501B开放模型。

Muse Spark Open ✨, Microsoft’s 3T param model 🧠, d1-ultra 💡

TLDR AI · 10/16 08:00

中文介绍 TLDR AI介绍Muse Spark Open、Microsoft的3T参数模型d1-ultra等AI技术。

Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint

Hugging Face Blog · 10/16 04:34

中文介绍 Hugging Face发布Granite Switch 4.1，实现从单一端点的聊天、决策和语音交互。

Bringing GPT-Rosalind to academic research labs

OpenAI News · 10/16 03:00

Introducing the Rosalind Academic Plan, offering eligible academic labs sponsored access to GPT-Rosalind and tools to advance life sciences research.

中文介绍 OpenAI推出Rosalind学术计划，为合格学术实验室提供GPT-Rosalind赞助访问和工具以推进生命科学研究。

Teaching students to think and learn through questions

OpenAI News · 10/15 21:00

Celebrate Teachers’ Day with AI for Educators in Brazilian Portuguese and see how Study Mode can help teachers support active, step-by-step learning.

中文介绍 OpenAI通过AI教育解决方案庆祝教师节，介绍Study Mode如何帮助教师支持主动学习。

[AINews] Modal’s $1B ARR, $16B Series D makes it NYC’s newest AI Infra Decacorn, 3x usage and 3x valuation in 5 months, 15x in 1 yr

Latent Space · 10/15 18:08

Sandboxes are all you need.

中文介绍 Modal实现1亿美元年度经常性收入和16亿美元的D轮融资，成为纽约市最新的人工智能基础设施独角兽公司。

Sullivan & Cromwell brings legal expertise into agreement review

OpenAI News · 10/15 08:00

In collaboration with OpenAI, Sullivan & Cromwell developed Agreement Analyzer, an AI tool for identifying issues in a draft agreement.

中文介绍 Sullivan & Cromwell与OpenAI合作开发协议分析器，AI工具用于识别草稿协议中的问题。

Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta

Hugging Face Blog · 10/15 08:00

中文介绍 Hugging Face推出Muse Spark Open，来自Meta的开放重量级多模态专家模型。

Welcome FLUX.3 Image Klein in Diffusers

Hugging Face Blog · 10/15 08:00

中文介绍 Hugging Face发布FLUX.3图像Klein，Diffusers中的新功能。

Jev 1.14 ⚡, Grok Voice 3 🎙️, RSI 🔁

TLDR AI · 10/15 08:00

中文介绍 TLDR AI介绍Jev 1.14、Grok Voice 3、RSI等功能更新。

Crossing the Threshold: How OpenAI and Anthropic Approach Autonomy and Safety in Financial Agents

Latent Space · 10/15 05:40

At AI Engineer New York, frontier labs and financial firms explored how to give agents useful autonomy while keeping consequential decisions under control.

中文介绍 AI工程师纽约会议探讨如何给予代理有用的自主权，同时控制重要决策。