Veille scientifique

Run AI Models for Edge Application Development with Just a Few Commands

Agent systems · primary_technical_publication

AI is expanding beyond the data center as developers build a new generation of agents to act at the edge: monitoring driver attention and cabin

From synthetic enterprise data evaluations to real Microsoft Copilot AI confidence

Evaluation & safety · primary_technical_publication

Evaluating Microsoft Copilot at enterprise scale requires test data that captures the full complexity of real organizations, yet real enterprise data contains sensitive information that

Build Specialized Agents on a Stack You Control with NVIDIA NemoClaw

Agent systems · primary_technical_publication

Building an autonomous agent requires a model, a harness, and domain-specific tools. To deploy an agent more securely, developers need a runtime that enforces...

Build a Validated Kubernetes Runtime for NVIDIA Dynamo

Compute & efficiency · primary_technical_publication

Teams building AI applications need a repeatable path from a model configuration to a working inference endpoint. NVIDIA Dynamo is a distributed inference...

Making it easier for websites to express AI training preferences

Data & governance · provider_announcement

OpenAI will honor a shared robots.txt signal to help website operators communicate their AI training preferences to participating providers.

Fair use benefits everyone

AI models · provider_announcement

OpenAI shares excerpts from court briefs by creators, academics, accessibility advocates, and legal experts, among others, on how fair use supports AI training and innovation.

nemotron-halloween

AI models · primary_technical_publication

Upcycling Language Models for Multiple Specialized Domains

AI models · preprint

Effective multi-domain adaptation of large language models requires prioritizing relevant domain knowledge while minimizing interference from irrelevant domains. Existing approaches commonly use Low-Rank Adaptation (LoRA)

TouchBridge: Tactile-Conditioned Human-to-Robot Transfer for Dexterous Manipulation

Robotics & embodiment · preprint

Human demonstrations offer an efficient and scalable source of data for dexterous robot manipulation, but differences in sensing and hand morphology complicate the transfer of

DynaRev: Benchmarking Web Agents under Dynamic Goal Revision

Agent systems · preprint

Web agents are typically evaluated with goals fully specified before execution and fixed thereafter, whereas real users often revise their requirements during interaction. To study

Granite Switch 4.1.1 — Update models like you update software

AI models · primary_technical_publication

RoboAtom: Large-scale Atomic-Action Supervision for Robot Learning

Robotics & embodiment · preprint

Fine-grained action annotations can support skill reuse and compositional generalization in robot policies. However, robot datasets typically provide task-level instructions without subtask annotations. Although some

Factorized Representations of Concurrent Temporal Variables in Pretrained Language Models

AI models · preprint

Language models often need to maintain multiple contextual variables that evolve independently within a document. Prior work has identified structured representations of individual temporal or

Pre-Crime for Agents, at the Last Second: Probing When a Recruited Agent Gives In

Agent systems · preprint

When AI agents share channels and credentials, one agent can press another to break its operator's rule, a first step towards collusion. Output filters check

Locating and Closing Information Gaps in Multi-Agent Systems

Agent systems · preprint

Multi-step Web tasks often require several LLM agents to gather and combine evidence. The communication graph of a fixed workflow restricts which agents can read

How Ropes & Gray is developing AI for transactional diligence

AI models · provider_announcement

Ropes & Gray is developing AI for transactional diligence with OpenAI. Lawyers review grouped documents, summaries, and reports before advising clients.

AI Cosmos: An Agentic Infra for Situated Learning

Agent systems · preprint

Large language models (LLMs) can improve task performance during learning, yet these gains do not necessarily reflect learners' ability to thereafter apply the knowledge and

Difficulty Calibration Across Time for Membership Inference in Diffusion Models

Compute & efficiency · preprint

Modern diffusion models are trained on massive datasets that often include copyrighted or private images. Membership inference attacks (MIAs) address this by testing whether a

Unlocking Tensor Core Performance for Machine Learning Interatomic Potentials

AI models · primary_technical_publication

Atomistic simulations help researchers understand how molecules and materials behave, from chemical reactions and phase transitions to mechanical failure. These...

Training language models in simulation for psychiatric intake

AI models · preprint

In psychiatric practice, the clinical intake interview remains a primary diagnostic instrument, requiring the clinician to establish rapport while disentangling complex, overlapping presentations. Here, we

Advancing Clinical Genomic Analysis with a Skill-Centred Agent

Agent systems · preprint

Clinical genomic analysis requires experts to integrate evolving guidelines, databases, and computational evidence. As sequencing has become increasingly accessible, this complex and labor-intensive process has

PhysVLMM: Interaction-Conditioned Physical Belief Maps for Articulated Objects in Robot Scene Maps

Robotics & embodiment · preprint

Mobile robots must open doors and drawers, yet their scene maps record what each object is and whether it has moved, not how it will

rTaste: Do Research Agents Have Taste? Evaluating Early Stage Research Judgment

Agent systems · preprint

Do research agents have taste, the judgment of which experiments are worth running before their outcomes are known? We give six frontier models seven recent

LLM-guided Planar Loco-Manipulation for Quadruped Robots

Robotics & embodiment · preprint

Robotic navigation typically assumes a collision-free route to the goal exists; in cluttered environments with movable obstacles, however, reaching the goal may require physical interaction.