Results for “Sam”

Daily Maverick Reuters

Taiwan, Japan flag incursions by Russian aircraft as Putin set to arrive in China

Taiwan said it detected two Russian military aircraft flying off its east coast on Tuesday, while Japan protested to Moscow after two aircraft believed to be Russian violated its airspace the same day.

Read at Daily Maverick

Android Authority Aamir Siddiqui Gadgets

Leak suggests Samsung delayed Quick Share’s Tap to Share for One UI 9.5

The "Tap to share" toggle only surfaces on Galaxy devices spoofed as running Samsung's unreleased software update.

Read at Android Authority

Analytics India Magazine Kileuna Haralu SI

AI Tools are Making Us Sound the Same at Work—Unless You're a Woman

A Johns Hopkins study finds differences in AI-generated professional writing across prompts with different linguistic features, prompting questions about how

Read at Analytics India Magazine

arXiv (research papers) Ivan Jaen Marquez SI

Learning Optimal Transport Plans Via Autoregressive Token Regression

Optimal Transport (OT) provides a principled framework for finding the minimum cost coupling between probability distributions. Classical solvers remain the gold standard for this task. At the same time, OT is an ideal testbed for studying whether learning-based amortized models can internalize the structure of OT type problems. Here, ground-truth solutions are available and marginal constraints are explicitly defined. Also, exact discrete solutions have well-characterized…

Read at arXiv (research papers)

arXiv (research papers) Ege Kaya University of British Columbia Security

Is 4-Bit Quantization More Than Noise to a Chat Model's Safety Behaviour? Attributing Behavioural Shifts with Matched Perturbations

Chat models on phones and laptops are usually 4-bit copies of the model that was evaluated, and audits of such copies disagree about what quantization does to refusal. A 4-bit build is one of many weight perturbations of the same size, so a change in its behaviour could come from its specific rounding error or from any perturbation that large. In a pre-registered study of three 3-4B chat models, with paired tests, blind human labels and an 8-bit null, we find two 4-bit…

Read at arXiv (research papers)

arXiv (research papers) Arjun Sivaraman SI

Readie: Low-Latency Code Sandboxes for AI Agents

From quantitative research to personal assistance, AI agents act by iteratively generating and executing code. To decouple the agent harness from code execution, many agents offload execution to dedicated systems called code sandboxes. However, existing code sandboxes suffer from slow startup. Whether they rely on custom snapshots or a single fat image, the root cause is the same: it is hard to build a compact, pre-warmed snapshot that can be shared across diverse…

Read at arXiv (research papers)

arXiv (research papers) Sushant Mehta SI

What RL on Expert-Built Tasks Teaches Nemotron 3.5 Super

We post-train NVIDIA's Nemotron 3.5 Super (120B total, 12B active parameters) with reinforcement learning (RL) on five expert-built task datasets, one at a time: real-world software engineering, complex constraints, long-horizon tool use, professional chart understanding and enterprise document reasoning. Each run trains a fresh rank-32 LoRA adapter with CISPO on Tinker, with no supervised stage, and each adapter is compared with the base model on the same tasks, harness and…

Read at arXiv (research papers)

arXiv (research papers) Allen G. Roush SI

Long-Context Generation Is a Sampling Problem (When You Turn Up the Heat)

Open-weight language models are widely observed to get worse the longer they generate. Often, attempts to fix this involve more training. At high temperature, The sampler, i.e. the rule that picks the next token out of the model's probabilities, decides whether long and agentic generation survives and/or thrives. We test 29 open-weight models from 1B to 122B parameters on ten established benchmarks plus needle retrieval out to one million tokens, covering SWE-bench Verified…

Read at arXiv (research papers)

The Guardian World Ben Smee Queensland state correspondent

‘Let me talk to Snez’: court hears recording of Sam Wood saying wife Snezana could recant domestic violence allegations

Fitness influencer’s lawyer calls comment during taped conversation between client and his father ‘ill-judged’ but denies it amounts to a suggestion of interfering with a witness Get our breaking news email, free app or daily news podcast A Queensland court has been played a prison recording of the former Bachelor star Sam Wood sobbing as he tells his father that he needs to talk to his wife, Snezana, and that his domestic violence charges would be dropped if she recanted…

Read at The Guardian World

arXiv (research papers) Rahim Hossain SI

LoCaR: Localized Canvas Retrieval With Frozen Vision-Language Model

Whole image embeddings can under represent small objects and local attributes that distinguish relevant images for localized queries. We present LoCaR, a training free framework for localized text to image retrieval with frozen vision language models. LoCaR constructs and utilizes independent proposals (SAM2) focusing on the object so as to get a more localized representation that independently encodes each region using the frozen visual encoder. We further study…

Read at arXiv (research papers)

Axios Barak Ravid Geopolitics

Top Trump aides to meet with Europeans, Ukrainians on peace plan

President Trump's envoys Steve Witkoff and Jared Kushner are planning to meet with Ukrainian and European officials this week to work on security guarantees and a postwar economic recovery plan for Ukraine, a U.S. official said. Why it matters: The Thursday meeting in Berlin comes a week after Witkoff and Kushner urged Russia to propose a postwar security framework for Europe — a parallel effort they hope will help bring the war to a close. On the same day, Trump will meet…

Read at Axios

arXiv (research papers) Qingcan Kang SI

RegimeCouncil: Paired-Evidence Routing For Memory Conflict Governance In LLM Agents

Language agents rely on long-term memory to personalize responses across extended interactions, yet stored facts quietly go stale as the world changes, and an agent often learns that using memory hurt only through delayed feedback after acting on an outdated record. Before the memory store is updated, how far should that negative feedback propagate: should it suppress memory use only on future queries for that same fact (a local policy), or also across other stored facts (a…

Read at arXiv (research papers)

Tech in Asia Minh Le Startups

Samsung Foundry to make Japan’s Preferred Networks AI chip

Preferred Networks said the L2000 series will succeed its MN-Core L1000 series, which is due in 2027.

Read at Tech in Asia

arXiv (research papers) Yifeng Peng

Commutator-guided repair and classical-accessibility boundaries for shallow quantum circuits

Trainability does not guarantee low-energy arrival: a shallow quantum circuit can be optimizable yet fail, under fixed depth and evaluation budgets, to reach Hamiltonian-selected states. We introduce a commutator-guided repair for this operational failure mode. Low-order host--Hamiltonian commutators and fixed same-support companion axes define a restricted pool of local Pauli rotations, from which a compact completion layer is selected before full reoptimization. Across 189…

Read at arXiv (research papers)

Quantum Computing Report Mohamed Abdel-Kareem

Quantinuum Details Nexus Hybrid Workflow Runtime Integration with Slurm and OpenQSE at SC26

Quantinuum's Nexus developer platform and Hybrid Workflow Runtime (Tierkreis) integrate quantum programming with supercomputing workload managers like Slurm, as demonstrated at SC26. This integration coordinates multi-stage scientific pipelines, preventing classical node idling while quantum jobs await processing. They also showcased a joint electronic-structure workflow with RIKEN, achieving lower ground-state energies for a 12-atom hydrogen chain through quantum-sampled…

Read at Quantum Computing Report

arXiv (research papers) Shinichi Morisaka

Analog-Memory-Based Waveform Generation for Qubit Control

We propose an analog-memory-based arbitrary waveform generator for cryogenic qubit control that separates low-speed waveform programming from high-speed waveform generation. Waveform samples are stored as capacitor voltages and sequentially accessed through buffered switching networks, enabling repeated, non-destructive readout without high-speed digital-to-analog conversion. A cryogenic prototype demonstrated Gaussian-shaped waveform generation at 14 K with a sampling rate…

Read at arXiv (research papers)

arXiv (research papers) Sadman Ahmed Shanto

QCap: a physics-corrected graph neural network for the capacitance matrix of any planar superconducting device layout

Electromagnetic simulations of device layouts to extract capacitance are computationally intensive, a major obstacle to accurate design of circuit devices. Machine learning models can bypass these simulations, but previous approaches have been restricted to layouts using only particular components, and require thousands of simulations of those same components as training data. We present QCap, a model that accurately predicts the Maxwell capacitance matrix of an arbitrary…

Read at arXiv (research papers)

Frontiers in Sustainable Food Systems Binapani Barik

Role of Spirulina and plant resins in sustainable edible packaging and smart food preservation systems

As the world goes circular, there is a growing need for sustainable substitutes of petroleum-based food packaging. The present article explores the potential of biopolymers produced from Spirulina platensis for facilitating the development of novel bioactive packaging materials. Due to their inherent antibacterial and antioxidant properties, the issues discussed herein address the challenges of minimizing plastic waste generation and improving food safety at the same time…

Read at Frontiers in Sustainable Food Systems

Simon Willison SI

llm 0.37.3

Release: llm 0.37.3 A bug fix release for LLM, mainly to cover a problem with the most recent Homebrew package of the software. Homebrew now ships the latest SQLite, 3.54.1, which includes a fix that broke LLM's logging code. Short version: in SQLite 3.54.0 the following SQL: select json_object('a', (select json_array(1))) Returns this: {"a":[1]} But in 3.54.1 the same query returns this: {"a":"[1]"} According to D. Richard Hipp: That was a bug fix. Per the documentation…

Read at Simon Willison

NASA Science Rafael Alanis Space

A Collage Showing 49 Drill Holes Made by NASA’s Curiosity Mars Rover

Description This collage shows 49 drill holes captured by NASA’s Curiosity Mars rover as of November 2026. Each hole represents a sample collected from rocks on the Martian surface; Curiosity’s drill, located on the end of its robotic arm, drops pulverized rock samples into instruments in its belly that analyze their composition. These images were

Read at NASA Science

Phys.org Quantum Physics

Measurements reveal link between quantum spin and sound-carrying vibrations in an ultrathin semiconductor

Researchers at the UCLA Samueli School of Engineering and Auburn University have achieved a scientific first: They measured atomic vibrations that carry heat and sound waves, and connected them to a quantum property called spin, which is a way of characterizing a subatomic particle's built-in magnetism.

Read at Phys.org Quantum Physics

arXiv (research papers) Tunahan Parlayıcı Security

Wrong Input, Wrong Threshold: Re-evaluating Adversarial Facial Privacy

Published face-privacy methods report protection above 90% against black-box recognition models. That number comes from an evaluation protocol with two independent defects, reported in a measure that cannot size either one. The input is wrong: methods published since 2022 feed the network the whole photograph, or a detector's box, where deployed systems detect, align to a landmark template and crop. Re-scoring the same files through a realistic front end costs a standard…

Read at arXiv (research papers)

GEN (Genetic Engineering & Biotechnology News) Sophia Ktori

Gut Microbome and Bile Acid Modification Linked with Longevity

Researchers studying how gut microbes modify bile acids have identified changes in the metabolic activity of the gut microbiome associated with exceptional longevity in humans, and found that amplifying the same activity with engineered bacteria improved metabolic health in aged mice.

Read at GEN (Genetic Engineering & Biotechnology News)

arXiv (research papers) Joshua B. Lovell Space

Rapid Protoplanetary Disk Evolution and Debris Disk Formation in Taurus-Auriga

We present a Submillimeter Array (SMA) 1.3mm (225GHz) continuum emission survey of Young Stellar Objects (YSOs) in the Taurus-Auriga star-forming region, comprising 69 class III and 5 class I/II YSOs with spectral types B9-M3, as part of a total sample of 149 class II and 82 class III YSOs. This represents a complete sample (100%) of class II and III YSOs with spectral types earlier than K3, and a near-complete sample (85%) for spectral types earlier than M2. The survey has…

Read at arXiv (research papers)

Kyiv Post Kyiv Post Geopolitics

Death Toll From Russian Strike Reaches 72 as Komyshuvakha Search Operation Concludes

Search and rescue operations concluded in Komyshuvakha, Zaporizhzhia region on Monday, Nov. 16, two weeks after Russian forces launched guided-bombs on a humanitarian aid site. According to Ukraine’s National Police, the death toll climbed to 72, with four people remaining missing. Russia shelled the village 11 times during the search, injuring police officers and rescuers. Regional Governor Mykhailo Semikin called it the deadliest attack in the region, while, on the same…

Read at Kyiv Post

The Verge Stevie Bonifield SI

Sam Altman says he wants everyone ‘to have a lot of RAM’

OpenAI CEO Sam Altman is once again addressing his controversial comments on PC RAM purchases. "I want to be clear, I want people to have a lot of RAM, I think everybody should be able to buy a lot of RAM for their own computer," Altman said today. "I think the home tinkering is great."

Read at The Verge

The Guardian US Guardian sport and agencies

USA thrash Canada to win women’s hockey worlds and complete historic golden double

USA beat Canada 6-2 to claim 12th world title Knight and Pannek score twice each in victory United States 6-2 Canada – as it happened The United States beat Canada 6-2 on Monday to win the women’s ice hockey world championship in Herning, Denmark, becoming the first US team to capture Olympic and world titles in the same calendar year. Hilary Knight and Kelly Pannek scored twice each as the Americans claimed their 12th world championship, adding to the Olympic gold medal…

Read at The Guardian US

Android Authority Stephen Schenck Gadgets

Samsung’s working on new international SIM tools for One UI 9.5

One UI 9.5 code shows travel SIM usage tracking by trip and day.

Read at Android Authority

arXiv (research papers) Ben Pennell Space

Brown dwarfs and stars are dynamically indistinguishable as companions in AU-separation binaries

Brown dwarfs occupy the mass range between stars and planets, populations whose dominant formation channels differ, which leave measurable differences in the orbital eccentricity and host-star metallicity distributions. We assemble a sample of astrometric binaries with main sequence primaries and brown dwarf companions at $\mathcal{O}$(AU) separations from $\textit{Gaia}$ DR3. The observable photocenter motions for such binaries are degenerate between a genuinely low-mass…

Read at arXiv (research papers)

Quantum Computing Report Mohamed Abdel-Kareem

IBM, RIKEN, and NVIDIA Achieve 79x Accuracy Gain in Quantum-Classical Simulations of Iron-Sulfur Clusters

Researchers from IBM Quantum, RIKEN, and NVIDIA have achieved a 79-fold improvement in energy accuracy for quantum-classical simulations of iron-sulfur clusters. This was done by initializing Sample-Based Quantum Diagonalization (SQD) with broken-symmetry unrestricted Hartree-Fock (BS-UHF) reference states, bringing ground-state energy estimates for a 72-qubit [Fe4S4(SCH3)4]2− active space within 4.3 millihartree of high-precision classical benchmarks. The study highlights…

Read at Quantum Computing Report

Older stories