Results for “models”

arXiv (research papers) Ji Guo Security

HalluProbe: Exploiting VLM Hallucinations for Backdoor Detection

Vision-Language Models (VLMs) are vulnerable to backdoor attacks introduced through data poisoning, making effective backdoor detection a critical security challenge. In this paper, we uncover a novel connection between VLM hallucinations and backdoor behaviors and exploit this connection for backdoor detection. Our key insight is that poisoned training data induces strong language priors toward the backdoor target token, while such priors are also a major source of…

Read at arXiv (research papers)

arXiv (research papers) Mi Luo

The Flip Test: Video World Models Fail to Learn the World State

Video world models promise to predict the future of a scene from its present. Do current models keep this promise? We probe this question counterfactually. When a small change to the present flips the future, such as a box resting a few millimeters before or past a table edge, a model that tracks the state of the world must flip its prediction too. We build FlipTest, a three-level hierarchy of increasingly strict tests over 1,680 initial frames from 10 simulated and 15…

Read at arXiv (research papers)

arXiv (research papers) Jiarui Dong

JEPA4R: Embodied JEPA World Models with 4D Point Planning

Joint-embedding predictive architecture (JEPA) world models predict future embeddings instead of pixels and plan in latent space. Two challenges remain: 2D images leave the metric geometry that manipulation acts on implicit and entangled with appearance, and what these models imagine stays hidden, as they are mostly judged in simulation by short-horizon latent planning. We present JEPA4R, a JEPA world model that observes single-depth-camera point clouds, predicts and plans…

Read at arXiv (research papers)

South China Morning Post Chong Ming Lee Geopolitics

Can Chinese AI companies curb cyber risks from open-weight models without losing their edge?

China’s leading artificial intelligence model companies are beginning to curb the cybersecurity capabilities of their most advanced open-weight models, following a string of high-profile AI hacking incidents involving both US and Chinese companies. Earlier this month, Chinese technology giant Tencent Holdings said it had stripped its latest Hy4 model of its cyber-hacking capabilities while retaining its cyber defence capabilities, following a similar move by Moonshot AI with…

Read at South China Morning Post

TechPowerUp DrUsmanSaleem

Newegg's RTX 5090 Restock Starts at $6,599, More Than Triple the MSRP

Newegg has restocked some GeForce RTX 5090 variants, but the cheapest RTX 5090 sold directly by Newegg at the time of writing is the GIGABYTE AORUS RTX 5090 Xtreme Waterforce at $6,599.99, about 3.3 times NVIDIA's $1,999 MSRP. The restock includes a handful of ASUS, GIGABYTE, and MSI models as first-party listings that ship from Newegg itself rather than from marketplace sellers. The AORUS Master ICE goes for $6,759.99, while MSI's Gaming Trio OC, Ventus 3X OC, and Suprim…

Read at TechPowerUp

Simon Willison SI

datasette-llm 0.1a11

Release: datasette-llm 0.1a11 datasette-llm is the intermediate plugin for Datasette that supports configuring LLMs for use with other Datasette plugins. This release adds support for OpenAI-compatible endpoints, so you can configure Datasette to use an LLM hosted anywhere that talks the OpenAI pseudo-standard web protocol. This means that if you're running local models with LM Studio or Ollama or llama.cpp (which just hit v1.0.0) you can run Datasette via uvx like this…

Read at Simon Willison

arXiv (research papers) Iaroslav Chelombitko SI

Do Rule-Based Tools Help LLMs Translate Endangered Uralic Languages in the Prompt or over MCP?

Komi-Zyrian, Moksha, Erzya and Udmurt have mature rule-based resources, among them finite-state analysers, constraint grammars and bilingual dictionaries, which language models do not use. We ask whether giving a model access to these resources improves machine translation from the four languages into English, and whether it matters how the access is provided: as analyses computed in advance and written into the prompt, or as tools that the model calls itself through the…

Read at arXiv (research papers)

The Decoder Matthias Bastian SI

OpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI unnecessary

OpenAI is dissolving its standalone robotics team for the second time, but not because it lacks data. General-purpose models like GPT-6 Astra can apparently control robots directly, making a dedicated unit unnecessary. The researchers are being redistributed across other teams in what the company frames as a way to accelerate its robotics efforts. The article OpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI…

Read at The Decoder

arXiv (research papers) Akshay Vyas SI

Neural Dual Bounds: Valid-by-Construction JGLP Warm-Starts for MAP and Constrained MAP

We present Neural Dual Bounds, a calibration-free framework that produces deterministic upper bounds for MAP inference in probabilistic graphical models using a single neural-network forward pass. The framework trains a network to output the edge messages of a mini-bucket style upper bound on a given join-graph decomposition in a self-supervised manner by directly minimizing a smooth dual objective. The output of this network can be used at test time to warm-start a standard…

Read at arXiv (research papers)

arXiv (research papers) Pengyu Chen SI

Mem2Fair: Online Fairness Control for Text-to-Video Generation via Fair Memory

Text-to-video (T2V) generation models can produce systematically unfair demographic outcomes when prompts leave protected attributes unspecified. For example, prompts describing nurses without specifying gender may disproportionately generate videos depicting female nurses. Mitigating unfairness in T2V generation is particularly challenging: training-based methods that modify model internals require architecture-specific access and risk disrupting temporal consistency, while…

Read at arXiv (research papers)

arXiv (research papers) Zane Fu SI

On-policy distillation as a Backdoor Propagation Channel

On-policy distillation (OPD) is emerging as an important paradigm for the post training of large language models, improving student models by leveraging supervision generated online by teacher models. However, whether this process also transfers latent security risks from teacher models to student models remains insufficiently studied. In this work, we reveal a new security risk in which an attacker first implants a conditional backdoor into the teacher model through…

Read at arXiv (research papers)

arXiv (research papers) Nikhil Kumar Jha SI

What Moves a LoCoMo Score? Memory Context, Answer Prompts and Judges

LoCoMo is a widely reported benchmark for long-term conversational memory, but published scores mix memory quality with choices of answer prompt, judge and baseline layout that are rarely reported together. We measure how much these choices move a score, and evaluate Becos, a team memory system for coding agents, as a black box under all of them: ten conversations (1,540 questions), two answer models (gpt-4o-mini, gpt-4.1-mini), two answer prompts and three LLM judges…

Read at arXiv (research papers)

arXiv (research papers) Jae Hun Hwang

Self-Calibrating Vision-Agent-Action: Deploying Pick-and-Place to New Robot Cells without Demonstrations or Fine-Tuning

In simulation, Vision-Agent-Action (VAA) performs pick-and-place in robot cells with gripper or base-frame changes without robot demonstrations or fine-tuning. It uses pre-trained vision-language and language models for interpretation, measures the command map and camera geometry from its own motion and fingertips, estimates lens distortion, and plans grasps from measured object depth. A controlled study of eight public LIBERO-fine-tuned vision-language-action models (VLAs)…

Read at arXiv (research papers)

arXiv (research papers) Kevin Nguyen SI

Isolating Perception for Multi-View Planning under Occlusion-Induced Partial Observability

Vision-language models (VLMs) are beginning to saturate visual planning benchmarks, yet they fail to resolve depth ambiguity and degrade sharply with problem size under occlusion-induced partial observability (PO). We introduce BW-PO based on the Blocksworld domain, a controlled benchmark that reduces existing VLM planners and formalizers to near-zero performance, and propose VLM-as-perceiver, a paradigm that narrows the VLM's role to local perception while delegating…

Read at arXiv (research papers)

arXiv (research papers) S M Rayeed SI

LABEL: Literature Asymmetry between Efficacy and Limits in Drug Knowledge and Language Models

Every medication comes with two types of medical facts: indications, which describe what conditions the drug treats (benefits), and contraindications, which describe who must avoid it to prevent serious harm (safety limits). Although contraindications are critical for patient safety and FDA regulations require listing them as rigorously as indications, we find that contraindications are far less documented than indications in general prose. In this paper, we measure how…

Read at arXiv (research papers)

arXiv (research papers) Rathindra Nath Karmakar

Asymptotic efficiency of tests for quantum statistical models

In classical statistics, local asymptotic normality provides a general framework for characterizing optimal testing power and constructing asymptotically efficient procedures. Extending this framework to quantum statistical models introduces additional challenges arising from the noncommutativity of observables and the need to optimize over quantum measurements. In this paper, we develop a theory of asymptotic efficiency for quantum hypothesis testing using quantum local…

Read at arXiv (research papers)

arXiv (research papers) Jieyuan Pei

SuperGlance: Driving at a Glance with a Vision-Language Model

Autonomous driving is turning to vision-language-action models (VLAs). They answer and plan by decoding, one step at a time. Each step is another forward pass, so a plan can take over a second of GPU time. Embodied AI often pairs such a slow, reasoning System 2 with a fast System 1 that acts. Human drivers plan at a glance, with System 1. Faster designs usually put System 1 outside the vision-language model (VLM), in a planner without its abilities. Yet before a VLM writes…

Read at arXiv (research papers)

arXiv (research papers) Cheongho Han Space

OGLE-2018-BLG-0829L: A Microlensing System with a Planet Orbiting a Star Possessing a Brown-Dwarf Companion

We analyze the microlensing event OGLE-2018-BLG-0829, whose light curve exhibits a weak short-term anomaly in addition to the prominent caustic-crossing features of a binary lens. The anomaly cannot be explained by a binary-lens single-source (2L1S) model. We therefore investigate both binary-lens binary-source (2L2S) and triple-lens single-source (3L1S) interpretations and find that both models reproduce the additional anomaly, with the 3L1S model being favored by…

Read at arXiv (research papers)

arXiv (research papers) Fatih Temiz

Wireless Edge-Assisted Adaptive Latent Reasoning for Vision-Language-Action Models in Physical AI

Vision-language-action (VLA) models enable embodied artificial intelligence (AI) by translating visual observations and language instructions into robotic actions. Emerging use cases in physical AI are expected to drive the need for efficient inference and communication, as the computational demands of the VLA models pose challenges for real-time control with limited onboard resources. Test-time scaling through latent reasoning allows these models to refine hidden…

Read at arXiv (research papers)

arXiv (research papers) Abhishek Rai SI

Beyond Plain MCP: Do Decision Models and Code Mode Make Tool Use Cheaper and Faster?

Model Context Protocol (MCP) gateways manage tool catalogs for language model agents using two primary paradigms: code mode where agents write scripts against tool stubs and decision layers which are cheap routers that preselect tool subsets. Using a controlled harness across 12 servers, up to 1,958 distractor tools, and three open-weight agents including Qwen, Gemma, and Nemotron, we present a preregistered evaluation on a sealed test split to determine when each approach…

Read at arXiv (research papers)

arXiv (research papers) Rui Zou SI

Agent-Guided Process Verification for Mathematical Self-Correction

Large language models have made substantial progress in mathematical reasoning, yet in complex multi-step reasoning, an error at any step may propagate along an incorrect path, and the resulting solution may omit necessary branches. Existing natural-language reflection methods often struggle to locate critical errors precisely, while outcome verification typically checks only whether the available candidate answers satisfy the problem conditions and can therefore overlook…

Read at arXiv (research papers)

arXiv (research papers) Nikhil Parkar SI

ST-GAT: A Leakage-Free Spatio-Temporal Graph Attention Framework for Robust Network Intrusion Detection

Intrusion detection systems are a central defence in modern networks. Deep learning models report strong anomaly-detection results, but evaluations on standard benchmarks such as CSE-CIC-IDS-2018 are often inflated by two methodological problems: temporal leakage from overlapping sliding windows, and contamination from preprocessing fitted before the train/test split. Per-flow classifiers also ignore the spatio-temporal dependencies across sequences of network flows. We…

Read at arXiv (research papers)

arXiv (research papers) Adriana Watson Security

When the Spec Sheet Lies: Injection Attacks on LLM-Generated Compliance Artifacts

As Large Language Models (LLMs) become used more commonly across a variety of fields, a growing body of research has highlighted particular concerns surrounding model security. Notably, prompt injection attacks, where malicious instructions are inserted into prompts, are particularly simple to implement yet capable of inducing devastating effects on the fidelity of model outputs. While there is substantial research surrounding attack models and their effects on chat-based…

Read at arXiv (research papers)

Tech in Asia Naomi Li Gan Startups

DeepSeek expands developer access with new AI APIs

Founded by Liang Wenfeng, DeepSeek already offered a public API and listed its available models.

Read at Tech in Asia

arXiv (research papers) Ran Qi

RoboCoEvo: Expanding Robot Capabilities through Agent-Policy Co-Evolution

Integrating MLLM-based agents with robotic foundation models has attracted increasing attention as a means of enhancing robotic capabilities. However, the improvement of such integrated systems remains largely decoupled across their agent and model components, with advances in one component not necessarily benefiting the other, substantially limiting the capabilities of the combined system. As both MLLMs and robotic foundation models become increasingly capable, fully…

Read at arXiv (research papers)

Tech in Asia Aiko Gao Ishida Startups

Samsung to produce DeepX’s next-generation AI chip

Aimed at robots and AI PCs, the DX-M2 is designed to run generative AI models on about 5 watts.

Read at Tech in Asia

Nhân Dân (Vietnamese state media)

Ha Noi enhances effectiveness of grassroots cultural and sports facilities

From reviewing existing conditions and revamping approaches to organising activities to developing models tailored to community needs, Ha Noi is focusing on boosting the effectiveness of its grassroots cultural and sports facilities, gradually transforming these venues into practical spaces for residents to gather and participate in regular events. Ha Noi to distribute 38,000 free tickets for concert celebrating Viet Nam Culture Day Ha Noi to stage special arts programme for…

Read at Nhân Dân (Vietnamese state media)

AWS Machine Learning Blog WonGu Rho SI

Production MLOps at GS Caltex with Amazon SageMaker Canvas, part 1

GS Caltex connects Amazon SageMaker Canvas to Amazon SageMaker Pipelines so process engineers' no-code models run in production. This post shows how ML engineers reuse validated preprocessing flows and models, cutting deployment from about a week to a few hours and running batch inference every 30 minutes.

Read at AWS Machine Learning Blog

arXiv (research papers) Asadullah Abdullah Khan Security

Agent-ZTA: A chain-Aware Zero Trust Model for Bidirectional Security-State Propagation and Runtime Containment in Multi-Agent Delegation

Multi-agent systems increasingly delegate tasks and permissions across chains of autonomous agents that invoke protected tools. Although agent identities and delegated credentials may remain valid throughout a workflow, the security condition of any agent along the chain can deteriorate during runtime execution-either through upstream compromise or downstream ingestion of untrusted inputs. Authorization models based solely on the immediate caller and its static delegated…

Read at arXiv (research papers)

arXiv (research papers) Rachel Hong Tech policy

Model Maps: A Documentation Framework for Tracing Foundation Model Ancestry and Understanding AI Entanglement

The development of foundation models incorporates pre-existing AI artifacts like models, datasets, and benchmarks. These relationships, as part of the broader AI supply chain, produce "entangled" webs in which properties of artifacts upstream can influence the outputs of a foundation model downstream - a phenomenon known as AI entanglement. Without visibility into these connections, a black-box model can inherit upstream properties that result in security vulnerabilities…

Read at arXiv (research papers)

Older stories