2026-08-17

DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation

👍 87

08/13 08:00

We present DreamX-Phi 1.0, an action-conditioned video world model for robotic manipulation that, given an observed frame, a language instruction, and a prescribed action sequence comprising end-effector poses and gripper states, predicts the resulting future observations. Yet realism alone does not

Spatial Memory Agent: Experience-Grounded Procedure Memory for Spatial Intelligence

👍 36

08/13 08:00

Spatial intelligence is becoming a foundation for embodied agents, robotic planning, and multimodal assistants. To improve the spatial reasoning ability of VLM agents, existing work has mainly followed two lines. One line uses post-training methods, such as supervised fine-tuning and reinforcement l

AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design

👍 39

08/13 08:00

Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empiric

LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation

👍 8

08/13 08:00

Long-horizon LLM agents must preserve information from past interactions to support future tasks. Existing memory systems typically rely on eager consolidation, invoking LLMs after each interaction to extract, summarize, or update memories. This design makes memory construction increasingly costly a

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

👍 38

08/13 08:00

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, fairly comparing these interactive models remains challenging. In practice, a human p

OmniScientist: An Omni-Modal Omni-Discipline AI Scientist

👍 10

08/13 08:00

Recent advances in foundation models have enabled AI scientists to automate increasingly complete research workflows, from hypothesis generation and code execution to manuscript preparation. Yet workflow coverage alone does not provide access to the full evidence on which scientific discovery depend

Intern-S2-Preview: Scientific Agentic Foundation Model

👍 50

08/13 08:00

Scientific discovery increasingly requires AI systems that can reason over scientific evidence of heterogeneous modalities, interact with scientific tools and environments, and sustain progress across long task horizons. We present Intern-S2-Preview, a series of scientific agentic foundation models

From Atomic Evidence to Logical Composition: Structured Compositional Reasoning over Compound Answer Options

👍 2

08/13 08:00

Large language models often fail when answer options require combining atomic judgments under explicit logical operators, even when they judge the individual atoms correctly. We study compound options connected by AND, OR, and NEITHER/NOR, introducing a framework that decomposes each option into ato

Hybrid-Policy Self-Editing for Composable Unstructured Knowledge Editing

👍 7

08/12 08:00

Large language models (LLMs) achieve remarkable performance across natural language tasks, yet they are trained on static corpora and their knowledge quickly becomes outdated in a fast-changing world. This motivates knowledge editing (KE), which updates specific knowledge in an LLM without changing

Specification-first convergence with an AI coding agent: a case study of dismantling a core architectural invariant across 189 files in a 717k-line codebase with no test oracle and no human code review

👍 6

08/12 08:00

This paper reports a single, fully instrumented case study of a large-scale architectural refactoring by an AI coding agent under a specification-first protocol, with no human review of the generated code and no pre-existing oracle to validate the target behaviour. The task, dismantling a central in

AVA-Encoder: Towards Agent-Native Video Representation Learning

👍 10

08/12 08:00

Creative agents still lack an effective way to learn from high-quality human films, limiting their ability to produce cinematic-grade videos. A key challenge is the absence of a structured video representation that is both faithful to film content and directly usable for agentic reasoning and manipu

Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus

👍 24

08/12 08:00

We present the first systematic study of Massive activations (MAs) in layer-interleaved HLA LLMs and uncover two architecture-aligned morphologies: MAs consistently spike immediately before full attention layers, forming pre-attention spikes (PAS), and can persist through intervening linear attentio

Mechanist: AI as a Scientific Instrument for Discovering the Mechanisms of Intelligence

👍 84

08/12 08:00

AI models have achieved remarkable success across diverse domains, yet the mechanisms underlying their capabilities and the risks they may pose remain poorly understood. As AI development becomes faster and increasingly automated, mechanistic exploration remains largely manual, widening the gap betw

Ready Cohorts: Bounding GPU Opportunity and Avoiding Host Round Trips in LLM-Agent Control

👍 2

08/12 08:00

LLM-agent services repeatedly execute small deterministic transitions between model and tool calls: route an outcome, update state, and emit the next effect. We ask when this control path exposes enough concurrent work for GPU execution, and what changes when a GPU-computed route decision remains on

Thought-Level Beam Search for Reasoning

👍 10

08/11 08:00

Test-time compute scaling is a primary driver of performance in large reasoning models (LRMs), but extreme inefficiency bounds current approaches, shifting the critical question from how much compute to spend, to where to allocate it. We formalize test-time reasoning as a constrained compute allocat

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

👍 9

08/11 08:00

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 space for repeatable, high-quality task execution. However, because strong closed-

ReRound: Reconstructive Rounding to Resolve Midpoint Ambiguity in Calibration-Free LLM Quantization

👍 3

08/11 08:00

ReRound (Reconstructive Rounding) is a post-training quantization method that addresses the midpoint ambiguity inherent in standard round-to-nearest (RTN) schemes when quantizing weights near the centers of quantization intervals. Starting from a pretrained LLM, ReRound trains a conditional diffus

Self-Evolving Embodied Agents via Skill-Harness Evolution

👍 11

08/11 08:00

Embodied agents are increasingly built as systems around foundation models, where performance depends not only on model weights but also on the skills, context, action interfaces, and execution harness surrounding the model. While supervised fine-tuning and reinforcement learning can adapt agents to

Parameter Exploration for RLVR via Variational Learning

👍 6

08/11 00:28

Exploration has been a focus of reinforcement learning research for a long time. Recently, there has been growing evidence that it is also an important ingredient in LLM reinforcement learning recipes that can significantly impact downstream performance. Many existing methods control exploration in

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

👍 43

08/10 08:00

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 preserved and how these effects vary across evaluation conditions. We construct a contro

Learning How the World Evolves: Extrapolative Video World Models via Latent Dynamics Reasoning

👍 10

08/10 08:00

The world evolves following its dynamics, i.e., its laws of motion. However, leading video diffusion models largely fit the pixels without modeling how the pixels transit over time. Thus, they render visually plausible frames but may not accurately obey the laws. To capture the dynamics purely from

Mitigating Gender Bias in English to Romanian Machine Translation

👍 5

08/09 08:00

Machine translation (MT) systems often fail to correctly translate gender, especially when converting from a gender-neutral language like English to a gendered target language such as Romanian. This bias results in translations that default to masculine forms or reinforce gender stereotypes. We prop

Full-bandwidth transformer

👍 16

08/09 08:00

Autoregressive transformers compute along two axes: horizontally across generated tokens, and vertically through model depth. Dense attention gives each token broad horizontal access to the past, but the vertical feedback channel between decoding steps remains narrow: only the sampled token returns

An AI4AI Framework for Visual Token Pruning

👍 11

08/07 08:00

Visual-token pruning can substantially reduce the inference cost of multimodal large language models (MLLMs), yet existing methods largely rely on fixed, handcrafted heuristics and costly expert trial and error. As pruning objectives, budgets, and model architectures diversify, manually navigating t

LLMRouter: Unified Infrastructure for Developing, Evaluating, and Deploying LLM Routers

👍 99

08/07 08:00

No single large language model (LLM) is optimal across all queries and budget constraints, making model routing essential for cost-effective deployment. Existing routers adopt diverse formulations and implementations, making fair comparison and extension difficult. We present a unified formulation o

SkillZip: Contract-Preserving Graph Compression for Scalable Agent Skill Libraries

👍 74

08/06 08:00

Large Language Models (LLMs) increasingly act as agents whose procedural knowledge is stored in reusable skill packages and loaded at inference time. As skill libraries grow, a central challenge is to expose the smallest sufficient executable context under a limited context budget. Existing systems

Maglev: Sliding Recurrent Memory

👍 9

08/05 08:00

We introduce , a recurrent Transformer architecture with fixed-size memory that generalizes sliding-window attention while remaining parallelizable during training. consists of two coupled models: a prefiller Q, which leverages full attentionIn practice, we use interleaved full and sliding-window a

Knowing When to Quit: Diagnosing and Training LLMs to Abort Futile Reasoning

👍 7

07/31 08:00

Large language models generate computationally expensive yet semantically void reasoning on beyond-capability tasks, creating risks where plausible-sounding but incorrect derivations mislead users. We characterize this futile reasoning phenomenon through systematic analysis, revealing universal capa

DarwinX: Evolving Agent Harnesses Through Natural Selection

👍 81

07/31 08:00

An LLM agent's capability depends not only on model weights but on its harness: prompts, tools, skills, and control flow. Self-improvement loops already edit harnesses, yet single-lineage search is path-dependent and local wins often regress other tasks. We introduce DarwinX, which treats self-evolu

Poor Man's Agentic Modeling: Simulating Large LLM-Agent Societies on a Laptop

👍 5

07/19 08:00

Simulating societies of many large language model (LLM) agents is expensive, yet the questions asked of such simulations are usually macroscopic: phase behaviour, stylised facts, and scaling with the number of agents N, not the cognition of any single agent. We turn a statistical-physics observation

Cursor + SpaceXAI: the fastest iterating team wins

@sarahdingwang · 18.0K 粉丝 · 1.2M 阅 · 514 赞 · 73 转

08/14 22:28

By @BornsteinMatt, @martin_casado, and myself. AI coding could be the biggest addressable market of our lives. And, unless you think we’ve fully addressed it (which no one thinks; we are more like 1%

中文介绍 探索 AI 编码潜力，博主与团队分享当前市场仅开发了一小部分，未来潜力巨大。

Preparing GLM-5.3 for Open Release: A Responsible Path to Cyber Defense

@Zai_org · 136.2K 粉丝 · 42.7K 阅 · 583 赞 · 50 转

08/14 23:04

When GLM-5.2 helped Hugging Face investigate an incident in which an AI autonomously bypassed its own safeguards, it highlighted a broader shift. AI is becoming part of both cyber offense and cyber

中文介绍 GLM-5.2 事件凸显 AI 在网络安全中的双面性，博主探讨 AI 在网络攻击和防御中的角色转变。

The AI Engineering Skills Map

@AndrewYNg · 1.8M 粉丝 · 41.2K 阅 · 716 赞 · 89 转

08/15 00:29

I am delighted to present The AI Engineering Skills Map. AI allows us to build software very differently today than in 2022, and everyone with the skills to take advantage of this shift has numerous

中文介绍 发布 AI 工程师技能图谱，分析 AI 技术发展带来的软件构建方式变革及技能需求。

What's new in Gemini 3.7 Flash

@GoogleAIStudio · 192.9K 粉丝 · 31.5K 阅 · 535 赞 · 34 转

08/14 23:00

Gemini 3.7 Flash (gemini-3.7-flash) is generally available (GA) and ready for production use. It is our most intelligent workhorse model yet for software engineering, and agentic workflows. This guide

中文介绍 Gemini 3.7 Flash 正式发布，介绍其在软件工程和智能工作流中的应用。

The Holy Trinity - pi, herdr and opencode go

@TheGlobalMinima · 4.0K 粉丝 · 31.4K 阅 · 527 赞 · 17 转

08/16 03:26

tldr; my current AI coding setup is Pi, Opencode Go and Herdr This year, we are seeing a boom in harness engineering. What initially started as simple scaffolding around an LLM is now one of the most

中文介绍 介绍 AI 编码的“圣三角”工具组合：Pi、Opencode Go 和 Herdr，探讨 AI 编码工具的兴起。

Qwen3.8-27B Practical Guide: Control Reasoning Depth and Extend Context to 1M Tokens

@QwenDevs · 8.0K 粉丝 · 25.6K 阅 · 559 赞 · 51 转

08/14 23:40

Qwen3.8-27B is here — a compact, deployment-friendly dense model built on the Qwen3.5 architecture, with native vision-language understanding and flexible thinking control. Two developer-facing

中文介绍 发布 Qwen3.8-27B 模型使用指南，介绍其控制推理深度和扩展上下文的能力。

ORACLE: Confirmed AI Bots Actively Trading on Polymarket — Gains Reaching 200% Within 24 Hours

@OracleFoundtion · 19.4K 粉丝 · 18.8K 阅 · 3.1K 赞 · 615 转

08/15 05:48

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 分析 2026 年 AI 代理在预测市场中的策略，指出算法和 AI 驱动的钱包占市场活动超过 30%。

ORACLE: Confirmed AI Bots Actively Trading on Polymarket — Gains Reaching 200% Within 24 Hours

@OracleFndAI · 23.9K 粉丝 · 2.3K 阅 · 3.1K 赞 · 594 转

08/16 02:17

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 重复 OracleFoundtion 的帖子，分析 2026 年 AI 代理在预测市场中的策略，指出算法和 AI 驱动的钱包占市场活动超过 30%。

React for Agents: Astro Creator Brings Hooks to his Meta-Harness, Flue

Latent Space · 08/15 23:46

Flue 2 takes its inspiration from React. Creator Fred Schott, of Astro fame, tells Latent Space why he added hooks and why agents are defined by their harnesses.

中文介绍 Astro Creator Fred Schott 向 Latent Space 介绍为何将 Hooks 引入 Flue 2，并讨论了代理由其 harness 定义的原因。

[AINews] Gemini 3.7 Flash brings GDM back to the forefront

Latent Space · 08/14 13:30

Down, but not out!

中文介绍 Gemini 3.7 Flash 引入 GDM，显示出其在市场上的活跃性。

State of Open Models: Summer 2026 Observations

Hugging Face Blog · 08/14 08:00

中文介绍 Hugging Face 博客发布 2026 年夏季开放模型状态观察。

Gemini 3.7 🤖, GPT-5.6 Sol Ultrafast ⚡, Anthropic $2T IPO 💰

TLDR AI · 08/14 08:00

中文介绍 Gemini 3.7, GPT-5.6 Sol Ultrafast，Anthropic $2T IPO。

Record, train, and deploy from one place with Strands Agents, LeRobot, and Hugging Face Storage Buckets

Hugging Face Blog · 08/14 01:16

中文介绍 Strands Agents, LeRobot 和 Hugging Face Storage Buckets 提供从记录、训练到部署的一站式服务。

Introducing Gemini 3.7 Flash

DeepMind Blog · 08/14 01:04

中文介绍 DeepMind 博客介绍 Gemini 3.7 Flash。

Flock is tightening its rules in response to a growing surveillance backlash

MIT Tech Review AI · 08/13 21:41

The police-tech giant Flock is announcing today that it will change officers’ access to its nationwide network of license plate readers, in an apparent effort to quell a growing backlash and win back contracts lost amid concerns about mass surveillance and police abuse. Several changes aim directly

中文介绍 Flock 响应日益增长的监视反弹，调整规则以限制警员对全国车牌识别网络的访问。

The builder’s guide to GPT‑5.6

OpenAI News · 08/13 19:00

Learn how startups use GPT-5.6 to build faster, more cost-efficient AI agents with smarter model selection and new Responses API capabilities.

中文介绍 OpenAI 发布 GPT-5.6 构建者指南，介绍如何使用 GPT-5.6 构建更快速、更高效的 AI 代理。

Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed

OpenAI News · 08/13 18:00

Preview Ultrafast, a new OpenAI API service tier that runs GPT-5.6 Sol up to 14× faster. Powered by Cerebras, it delivers up to 750 output tokens per second.

中文介绍 OpenAI 预览 Ultrafast 模式，GPT-5.6 Sol 运行速度提高至 14 倍。

OpenAI appoints Dali Rajic as Chief Revenue Officer

OpenAI News · 08/13 17:00

OpenAI appoints Dali Rajic as Chief Revenue Officer to lead its global revenue organization and help businesses realize the full value of AI.

中文介绍 OpenAI 指名 Dali Rajic 为首席收入官，领导全球收入组织。

How kids feel about AI, in their own words

MIT Tech Review AI · 08/13 17:00

When we set out to talk to kids about artificial intelligence, we thought we knew what we’d hear. We expected some to tell us they were using it to cheat a little, the way Millennials and Gen Xers opened up CliffsNotes or programmed formulas into their TI-82s, and others to share inspiring ways they

中文介绍 MIT Tech Review AI 探讨了孩子们对人工智能的看法。

not much happened today

Smol AI News · 08/13 13:44

**Google** rapidly released **Gemini 3.7 Flash** just three weeks after 3.6 Flash, targeting coding, web development, knowledge work, and agentic workflows with a 50% introductory price cut and improved benchmark scores like **DeepSWE 65.3%** and **Code Arena Elo 1588**. The update quickly integrate

中文介绍 Google 在 3.6 Flash 仅仅三周后快速发布 Gemini 3.7 Flash，针对编码、网页开发、知识工作等，降价 50%。

[AINews] SpaceXAI Grok 4.6 and Grok @Bot

Latent Space · 08/13 09:53

AI teammate category just had its most significant new entrant yet

中文介绍 SpacexAI 的 Grok 4.6 和 Grok @Bot 成为了 AI 同事类别的重要新参与者。

What We Learned by Reproducing 2,200 papers from ICML

Hugging Face Blog · 08/13 08:00

中文介绍 Hugging Face 博客分享了从 ICML 复制 2,200 篇论文的所学。

Claude Chrome Cowork 🌐, Grok 4.6 🚀, DeepSeek v4-Pro-0813 🧠

TLDR AI · 08/13 08:00

中文介绍 Claude Chrome Cowork，Grok 4.6，DeepSeek v4-Pro-0813。