2026-10-12

The Lattice of Transition Laws

👍 5

10/08 08:00

Diffusion and autoregression (AR) have long been seen as different categories of generative models, with diffusion specialising in continuous fields and AR specialising in discrete tokens. Recent work seeks to combine the advantages of the two models, and each hybrid fixes its decoding schedule by d

Opera: A Verbal Critic Framework for Long-horizon Coding Agents

👍 14

10/08 08:00

Long-horizon coding agents need timely corrections, yet feedback can be ineffective or even harmful when it misjudges ongoing work or fails to address the underlying problem. Existing critics focus on evaluating trajectories and generating feedback, but rarely track what happens after feedback is de

Skill Constellations: Tracing the Supply Chain of Agent Skills on GitHub

👍 5

10/08 08:00

Agent skills are SKILL.md instructions and scripts that AI coding agents such as Claude Code and Codex run with the permissions of their user. Developers share skills by copying them between repositories, which makes them a software supply chain without a registry, versions or provenance. The origin

One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts

👍 7

10/08 08:00

In this work, we show that a single Transformer block, applied recurrently, can match the accuracy of a full-depth vision encoder at comparable inference FLOPs without intermediate feature distillation. reViT restores depth-specific transformations by representing the FFN at each recurrent depth as

A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box Optimization

👍 8

10/08 08:00

Black-box optimization (BBO) arises in many scientific and engineering problems where objective evaluations are expensive and limited. Recent large language model (LLM) agents offer a new way to approach BBO by combining task semantics, computation, optimization tools, and feedback-driven decision m

Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks

👍 35

10/08 08:00

Learning to act in unfamiliar environments requires agents to infer how the world works and revise that understanding as new evidence arrives. Yet limited observations can support multiple world models that explain past interactions but predict different outcomes in unseen states. We introduce Memen

REMORY: Learning Residual Memory for Context Compaction

👍 38

10/08 08:00

Long-horizon agents compact their history to continue within a finite context window, but a textual summary alone may not support every subsequent decision. We introduce REMORY, a neural memory network that supplements the summary with a bounded sequence of soft memory tokens. Given the history and

MARGIN: Runtime Confidence Calibration for Multi-Agent Foundation Model Coordination

👍 5

10/08 08:00

When a coordinator compares answers from heterogeneous foundation models, self-reported confidence may have different meanings across responders and changing workloads. This paper presents MARGIN (Multi-Agent Runtime Grading via Incremental Normalisation), a runtime calibration method that learns mo

MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement

👍 74

10/08 08:00

Reinforcement learning (RL) is the central training paradigm for advancing large foundation models towards self-improvement. This report introduces the MiMo-V2.6 series, an omni-modal family that pushes the frontier of model intelligence by scaling RL compute. Prior to RL, we conduct mid-training on

Foundations of Large Language Models

👍 42

10/08 08:00

This is a book about large language models. As indicated by the title, it primarily focuses on foundational concepts rather than comprehensive coverage of all cutting-edge technologies. The book is structured into six main chapters, each exploring a key area: pre-training, generative models, prompti

BrickBench: Evaluating Agentic Brick Design

👍 7

10/08 08:00

We propose BrickBench, a benchmark for agentic text-conditioned LEGO-set design. Given a prompt, an agent is tasked with producing an assembly that not only satisfies semantic and design criteria, but that can also be physically built. To do so, it must select parts from a discrete library and reaso

OmniCapBench: A Deep-Structured Evaluation Framework for Fine-Grained Audio-Visual Captioning

👍 15

10/08 08:00

Multimodal large language models (MLLMs) are rapidly evolving toward continuous audio--visual reasoning, creating an urgent need for evaluations that expose their capability limits. Audio--visual captioning is an ideal diagnostic task, yet current benchmarks face a coupled trade-off: whole-caption s

Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict

👍 8

10/08 08:00

When retrieved evidence contradicts an agent's prior beliefs, does it revise its answer, acknowledge uncertainty, or persist with an incorrect conclusion? Existing evaluations of agentic systems focus primarily on task success, offering limited insight into how agents handle such conflicts. We propo

Evaluating the Transfer of Co-Evolved Communication from 2D to 3D Simulation

👍 4

10/07 08:00

This work examines the transfer of a co-evolved communication mechanism between two robotic agents from a discrete two-dimensional (2D) simulator to a three-dimensional simulator with real physics (3D). The study focuses on whether a communication mechanism co-evolved in a 2D environment retains its

MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent

👍 8

10/07 08:00

Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user intent. A global text-audio relevance score can overlook the implicit intent in underspecified prompts and mask failures in specific requirements, such as instrumentation,

On-Policy Distillation Teaches New Skills but Not New Knowledge

👍 6

10/07 08:00

On-policy distillation (OPD) strengthens language-model reasoning, yet whether students acquire new factual knowledge or compositional skill for multi-step reasoning remains unknown. We separate these capabilities using a controlled synthetic framework that measures the student's initial capabilitie

Incidental information contaminates patient notes and disrupts clinical reasoning in large language models

👍 6

10/06 08:00

Large language models (LLMs) are increasingly relied upon to support ambient documentation and clinical reasoning. Here we examine the impact of a failure mode shared between these two applications by assessing their sensitivity to information incidental to the patient encounter. In 576 patient-clin

Frozen Models, Evolving Expertise: Model-Agnostic Learning from Deployment Experience for Multimodal Medical AI

👍 6

10/06 08:00

Large language models (LLMs) and vision-language models (VLMs) are usually frozen after deployment, so they do not learn from the cases they solve. This is especially concerning in medicine, where new clinical evidence, updated guidelines, and new therapies can change established practice. Fine-tuni

SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Models

👍 7

10/06 08:00

Diffusion large language models (DLLMs) generate text through iterative block denoising, and multi-branch speculative decoding accelerates this process by verifying a main branch together with multiple draft branches in a single forward pass. While prior DLLM acceleration methods primarily exploit t

Can AI Agents Make Open-Ended Scientific Discovery? Evidence from Station

👍 8

10/06 08:00

Recent AI systems have made rapid progress in scientific discovery when given well-defined metrics, but whether they can autonomously undertake open-ended scientific discovery remains unclear. We investigate AI's ability to tackle open-ended tasks in Station, an open-world environment in which multi

EDiS: Edge Disjoint Subgraph Sparsification Framework for Graph Neural Networks

👍 6

10/06 08:00

Sparse GNN training reduces computation, but deciding which edges to keep can be costly. Reusing one sparse graph is cheap, but locks training to a fixed topology, while varying it across epochs can require repeated sampling or recomputation. We introduce EDiS (Edge-Disjoint Subgraph sparsification

U-Space: Uncovering When and Why Uncertainty Arises in Language Models

👍 39

10/06 08:00

Large language models are informing decisions with ever-higher stakes. As the consequences of their errors grow, a central question becomes harder to ignore: how much can we trust an individual answer? Yet recognizing when to defer remains difficult because language models can present incorrect conc

CARE: Certifying Acceleration for Vision-Language-Action Inference

👍 7

10/06 08:00

While vision-language-action (VLA) models have advanced rapidly, running them at every control step remains expensive. Prior work accelerates VLA inference using techniques like action chunking and visual-token pruning, typically evaluating based on latency and average task success. However, acceler

SPW-Nav: A Streaming Panoramic World Model for Language-Guided Navigation

👍 5

10/06 08:00

Language-guided panoramic video generation benefits various downstream applications, such as interactive 3D scene exploration, virtual reality experiences, and embodied agent training. Existing panoramic generators follow predefined trajectories, and interactive world models act through low-level ac

Investigating the Role of Reasoning-Language Alignment in Monolingual Retrieval-Augmented Generation

👍 5

10/02 08:00

Reasoning traces improve large language models (LLMs), but current models are trained to reason mostly in English. It has been shown that forcing a model to reason in another language degrades accuracy, even when the reasoning language matches the language of the prompt -- but only for a setting whe

Behavioral Persistence and Incomplete Functional Transfer of Co-evolved Communication in Evolutionary Robotics

👍 4

09/29 08:00

This work evaluates the direct transfer of a co-evolved communication protocol from a 2D simulation to a 3D physical environment, without retraining the network weights. Two e-puck-type robots, controlled by a GRU network with residual connection, were evaluated in a food-seeking task with social si

Learning to Steer, Steering to See: Unveiling the Geometry of RLVR in Large Language Models via Trainable Vectors

👍 5

09/28 08:00

Reinforcement learning (RL) has become a key paradigm for enhancing the reasoning of large language models, yet the high dimensionality of parameter updates makes its training dynamics hard to analyze. We study reinforcement learning with verifiable rewards (RLVR) and use vector steering to identify

ReSPO: Reshaped Sequence Policy Optimization for Gradient Starvation in Off-Policy Learning

👍 12

09/28 08:00

Reinforcement learning from verifiable rewards (RLVR) frequently reuses rollouts across multiple policy updates, increasing the mismatch between the current policy and the data-generating policy. We identify a sign-dependent gradient starvation problem in clipped policy optimization: clipping suppre

SatNav: A Scalable Benchmark for Long-Horizon UAV Vision-Language Navigation from Satellite Imagery

👍 5

09/25 08:00

Urban uncrewed aerial vehicle (UAV) vision-language navigation (VLN) requires agents to follow instructions across extended urban spaces, inherently demanding long-term memory and geospatial grounding. However, scaling existing benchmarks remains difficult because of their reliance on costly reconst

Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction

👍 7

09/24 08:00

Tool-calling agents have become central to enterprise AI, yet training and evaluating them at scale remains severely constrained due to business and legal restrictions on enterprise systems, data, and database schemas. Tabular data synthesis offers a natural alternative, but its effectiveness is fun

Models as Insider Risks in the Super Intelligence Era

@satyanadella · 9.6M 粉丝 · 448.2K 阅 · 1.3K 赞 · 183 转

10/10 22:43

As traditional software systems were being deployed across the economy over the last few decades, we had the tools and capability to trace behaviors to a specific code path. That same kind of

中文介绍 纳德拉探讨超级智能时代模型作为内部风险的问题，指出传统软件系统追踪行为的能力在智能时代面临挑战。

Agentic evals: how to know whether an AI agent did the job, and will do it again

@sermakarevich · 1.0K 粉丝 · 30.6K 阅 · 504 赞 · 80 转

10/10 23:25

An article for engineers who ship agents built on large language models (LLMs): systems that call tools, change records, and act on a customer's behalf over several steps. After reading it you will be

中文介绍 瑟梅卡雷维奇分享工程师如何评估基于大型语言模型的AI代理是否完成工作及其可靠性，提供实用评估方法。

[Subscriber Exclusive] NYC Subscriber Meetups!

Latent Space · 10/11 06:07

Tomorrow and Tuesday. If you see this you’re in!

中文介绍 Latent Space将举办纽约订阅者聚会，时间为明天和周二。

Building AI for Reliable Execution: Lessons From Industrial Robotics

Latent Space · 10/10 22:04

Inside Standard Bots’ AI stack, pretrained models learn factory tasks from demonstrations and improve through corrections from real deployments.

中文介绍 Standard Bots展示了其AI堆栈，通过演示和实际部署中的纠正来学习工厂任务。

[AINews] TypeSafe/Jev at >$100M ARR, $7.5B valuation 3 weeks after launch

Latent Space · 10/10 14:45

Wow.

中文介绍 TypeSafe/Jev在推出3周后达到超过1亿美元的收入和75亿美元的估值。

Why AlphaFold Didn't Solve Protein Folding — Pushmeet Kohli, Google DeepMind & Sal Candido, Biohub

Latent Space · 10/10 08:31

From the Bitter Lesson of AI scaling to the unsolved mysteries of protein folding, Google DeepMind’s Pushmeet Kohli and Biohub’s Sal Candido are rethinking what it takes to build AI that truly understands biology.

中文介绍 Google DeepMind的Pushmeet Kohli和Biohub的Sal Candido正在重新思考构建真正理解生物学的AI所需条件。

Impactful scheduling for GPU clusters

Hugging Face Blog · 10/09 23:20

中文介绍 Hugging Face博客讨论了GPU集群的有效调度。

We’re putting too much faith in AI’s ability to say no

MIT Tech Review AI · 10/09 17:00

Ever since people first seriously contemplated giving machines an intelligence modeled on our own, there has never been any question that they would, like us, be able to say no. The sci-fi canon is full of stories of robotic disobedience. Most of these capers are, of course, cautionary. But recently

中文介绍 MIT Tech Review AI指出，人们对AI拒绝能力的信任过多。

Sophos cuts threat investigation time by 96% with OpenAI Daybreak

OpenAI News · 10/09 15:00

Discover how Sophos uses OpenAI’s Daybreak to cut cyber-threat investigation time by 96% and automate 52% of MDR cases while preserving human oversight.

中文介绍 Sophos利用OpenAI的Daybreak将网络安全威胁调查时间缩短96%。

Asana cuts model costs 76x in browser tests with GPT-6.1 Sol

OpenAI News · 10/09 15:00

Using GPT-6 Astra in Codex, Asana made its browser agent 76x cheaper and 5x faster in tests to offer customers more capable models.

中文介绍 Asana使用GPT-6 Astra将浏览器代理的成本降低了76倍。

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

MIT Tech Review AI · 10/09 08:08

Friday, October 16, 2026 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 could be next

中文介绍 MIT Tech Review AI将举行关于AI设计病毒创造者的圆桌讨论。

GPT-6.1 Ultrafast ⚡, Gemini universal agent 🤖, speculative decoding 🧠

TLDR AI · 10/09 08:00

中文介绍 TLDR AI介绍了GPT-6.1 Ultrafast、Gemini universal agent和speculative decoding。

[AINews] not much happened today

Latent Space · 10/09 07:29

a quiet day.

中文介绍 Latent Space报道今天没有发生太多事情。

Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk

Latent Space · 10/09 00:27

A special Science pod and Engineering pod crossover.. with Forward Deployed Engineering kicker!

中文介绍 Periodic Labs的Liam Fedus和Ekin Dogus Cubuk探讨了从半导体到超导体的合成超级智能。

How Oracle turns days of work into minutes with ChatGPT and Codex

OpenAI News · 10/09 00:00

Across recruiting, engineering, and operations, Oracle turns specialist knowledge into fast, repeatable workflows with ChatGPT Work and Codex.

中文介绍 Oracle利用ChatGPT和Codex将工作日缩短到分钟。

Pollo AI turns creative ideas into campaigns with OpenAI

OpenAI News · 10/08 20:00

With GPT-5.6, GPT-6 Astra, and GPT‑Image‑2.5, Pollo AI helps creators turn bold ideas into detailed images and cinematic video ads.

中文介绍 Pollo AI利用OpenAI将创意想法转化为详细图像和视频广告。

LegalOn halves Codex costs while maintaining development speed

OpenAI News · 10/08 20:00

LegalOn cut estimated daily Codex costs by 65% while maintaining development speed. It matched Astra, Sol, and Luna to tasks and managed budgets strategically.

中文介绍 LegalOn将Codex成本削减了一半，同时保持开发速度。