Results for “model”

arXiv (research papers) Zeming Yu SI

Stop Going in Circles:Loop-Triggered Routing for System 1 Decision Agents

System 1 decision models (represented by Jev) provide fast, low-cost action selection, but existing agent integrations do not reliably translate these local advantages into task completion. At certain points during execution, an agent may repeatedly select the same actions without making progress toward its goal. LLM-based policies offer strong reasoning capabilities that support planning within an agent, but invoking them at every step incurs substantial overhead. The…

Read at arXiv (research papers)

arXiv (research papers) Bochen Han SI

Incorporating argumentation frameworks for LLM decision-making: Task-specific design and evaluation

Argumentation frameworks (AFs) provide explicit, formal mechanisms for representing and resolving conflicts and thus have been incorporated into large language model (LLM) decision-making. The benefits vary and depend on how arguments and their relations are constructed, evaluated, and incorporated into the decision process. {We abstract a workflow consisting of four stages based on existing argumentation-augmented methods} and investigate their design choices over tasks of…

Read at arXiv (research papers)

YourStory Startups

Debate brews over Sarvam hosting Chinese AI models; Grand Theft Auto VI is finally here

Token Factory is an inference platform, which provides the computing infrastructure required to run trained AI models and generate responses.

Read at YourStory

arXiv (research papers) Sanjari Srivastava SI

AppSci-Bench: Can LLMs Build AI Agents and Make Them Better?

Large language models (LLMs) are increasingly used not only to power AI agents but also to build them. While many benchmarks assess the quality of LLM- powered agents, few ascertain how good LLMs are at building them. To this end, we introduce AppSci-Bench, where we construct a setting akin to how human Applied Scientists build agents for businesses - through eliciting task requirements from stakeholders, exploring the business's infrastructure, and scientifically improve…

Read at arXiv (research papers)

Tech in Asia Grace Priscilla Teo Startups

Palo Alto Networks to buy AI firm Applied Compute for $4.5b

Applied Compute sells software that helps companies adapt and run these models on their own data.

Read at Tech in Asia

Tech in Asia Naomi Li Gan Startups

Google rolls out Gemini 4 Neon AI model for coding, science

It is available through the Gemini API in Google AI Studio, Gemini for business, and the Gemini app for Pro and Ultra subscribers.

Read at Tech in Asia

arXiv (research papers) Yixiong Chen SI

Pro$^2$ Agent: Scientific Problem Probing for Autonomous AI R&D

Large language model (LLM) agents can now autonomously conduct AI research by writing code and running experiments. They typically do so by iteratively optimizing a codebase, while ignoring the problem structure behind the task. This formulation is effective on simpler tasks, but on difficult problems it often gets stuck in the wrong solution region because the agent never explicitly represents its understanding of the problem. We introduce Pro$^2$ Agent, which shifts the…

Read at arXiv (research papers)

Tech in Asia Grace Priscilla Teo Startups

Microsoft’s reported ByteDance AI deal draws House scrutiny

The reported US$500 million deal would let ByteDance access OpenAI models.

Read at Tech in Asia

Tech in Asia Grace Priscilla Teo Startups

Google rolls out Genie 4 world model to AI Ultra users

Google said generated video and audio carry SynthID watermarks, and all model inputs and outputs undergo safety screening.

Read at Tech in Asia

Viet Nam News (state-run daily)

Integrated industrial-energy centre model critical for Việt Nam to drive growth

The key is to move away from fragmented development towards shared infrastructure, coordinated investment and complementary value chains.

Read at Viet Nam News (state-run daily)

arXiv (research papers) Ahmet Efe Security

Verified Netlist-to-RTL Recovery with LLM Agents

Gate-level netlists circulate among IP integrators and foundries and can be extracted from chips. Reverse engineering them into readable register-transfer-level (RTL) code would help an adversary understand and modify a design, but remains difficult because word-level operations are scattered across many gates. Large language models (LLMs) can write and optimize RTL, but often fail to recover it from a whole netlist. We propose LOCA (LOgic Cone Attack), an LLM-based attack…

Read at arXiv (research papers)

The New Stack Amanda Caswell

Google’s Gemini 4 Neon tripled its coding score without changing the model

Google’s new Gemini 4 Neon resolved 12.8% of the SWE-sweep coding benchmark using the mini-SWE harness, but that score jumped

Read at The New Stack

Phys.org Space Space

First lunar-specific ground motion model could guide seismic hazard assessments on moon

Civil engineers looking forward to building structures on the moon need to know how ground motion from moonquakes will affect these buildings. But will ground motion models developed for Earth work well for the lunar landscape?

Read at Phys.org Space

AWS Machine Learning Blog Sandeep Raveesh-Babu SI

Full Fine-Tuning (FFT) on SageMaker AI Serverless Model Customization

Amazon SageMaker AI Serverless Model Customization now supports Full Fine-Tuning (FFT). This post explores when FFT outperforms LoRA across key benchmarks, examines the spectral properties of model weight matrices behind the quality gap, shares practical hyperparameter tuning guidelines, and shows how to get started.

Read at AWS Machine Learning Blog

Graham Cluley Graham Cluley Security

Smashing Security podcast #493: Why your AI model could be a sleeper agent

You join a video call, and your contact and a few colleagues are already there, waiting patiently. But they can't hear you, and a pop-up says you need to install an update to fix your microphone. What could possibly go wrong? Meanwhile, we explain how changing just a handful of numbers can turn an AI model into a sleeper agent, waiting for the right moment to strike. Plus, Andrius Buinovskis of NordLayer explains how companies can see which AI tools their staff are really…

Read at Graham Cluley

SiliconANGLE AI Maria Deutscher SI

Google upgrades Gemini’s agentic Task mode with new model, e-commerce features

Google LLC is releasing a new version of its Gemini artificial intelligence assistant that is better at performing multistep tasks. Josh Woodward, the vice president responsible for the Gemini app, detailed the update in a Tuesday blog post. The AI assistant has two modes that are accessible through separate interface tabs. The first, Chat, generates

Read at SiliconANGLE AI

Frontiers in Sustainable Food Systems Mingzhi Che

Industrial specialization and agricultural green production efficiency: local effects and spatial spillovers

Cross-sector linkages between industrial specialization and agricultural green production efficiency, and their spatial distribution, are important for understanding coordinated industrial and agricultural transformation. Using 3,432 observations from 264 Chinese cities during 2011–2023, we examine local and spatial associations with two-way fixed-effects and spatial Durbin models. Efficiency is measured using a global, non-oriented slacks-based measure (SBM) under variable…

Read at Frontiers in Sustainable Food Systems

Frontiers in Sustainable Food Systems Limin Cheng

Regional artificial intelligence development and agricultural ecological efficiency: evidence from Chinese provinces

Existing studies have primarily focused on the effects of artificial intelligence (AI) on productivity and carbon emission efficiency, while systematic evidence on whether and through which mechanisms AI improves agricultural ecological efficiency (AEE) remains relatively limited. Using provincial panel data from China for the period 2005 to 2023, this study employs the entropy weighting method to measure regional AI development and a slack-based measure (SBM) model…

Read at Frontiers in Sustainable Food Systems

TechCrunch Sarah Perez SI

Satya Nadella predicts AI agents will reshuffle the economics of the web

Microsoft's CEO says AI agents could disrupt advertising-driven business models, forcing companies like Amazon to rethink how they compete for customers.

Read at TechCrunch

arXiv (research papers) Tunazzina Islam Tech policy

When Objectives and Evaluators Disagree: Evaluating Transfer for Post-hoc Demographic Framing Mitigation

Post-hoc rewriting is used to reduce demographic disparities in LLM-generated text when the source model is inaccessible. A rewrite can look successful merely because it improved its own target or satisfied its selector. We study this as a transfer problem on 660 gender pairs and 1,980 age contrasts of demographic-conditioned climate messages. A rewriter proposes edits, a selection model chooses among constraint-satisfying candidates, and an independent evaluator uninvolved…

Read at arXiv (research papers)

AWS Machine Learning Blog Dan Ferguson SI

Introducing the Amazon SageMaker JumpStart Model Leaderboard

The Amazon SageMaker JumpStart Model Leaderboard compares 70+ foundation models side by side on quality, speed, and cost, directly in SageMaker Studio. Learn what the leaderboard shows, how AWS benchmarks models, and how to go from a shortlist to a deployed model.

Read at AWS Machine Learning Blog

arXiv (research papers) Yihan Wang Security

Reject and Report: Forgery-Evident Undetectable Watermarks for Generative Image Models

Provenance schemes such as those based on content watermarks let images from a generative model be traced back to it. A practical watermarking scheme must be accurate at a low false-positive rate (FPR), robust under content-preserving processing, and undetectable. In addition, it must resist forgery; otherwise, even from a single watermarked image and a surrogate model, an adversary can imprint the watermark onto an arbitrary image or regenerate new content that carries it…

Read at arXiv (research papers)

AWS Machine Learning Blog Alan Yao SI

How Airbnb uses cost per task to operate coding agents on Amazon Bedrock

Airbnb runs evals on real engineering tasks and uses Amazon Bedrock observability to choose models for its coding agents and control production costs. This post shows how cost per task guides model selection, why production spending can differ from eval results, and how to monitor latency, manage quotas, and attribute spend.

Read at AWS Machine Learning Blog

9to5Google Abner Li SI

Google launches Gemini 4 Neon with Carbon quality at Flash speeds, live for AI Pro

Following last week’s release of Carbon and Hydrogen, Google today is rolling out Gemini 4 Neon. The model offers “frontier-level reasoning for complex, high-frequency workflows faster and more cost efficiently.”

Read at 9to5Google

GEN (Genetic Engineering & Biotechnology News) Julianna LeMieux, PhD

Targeting Folate Receptor Reverses Colorectal Cancer’s Resistance to Immunotherapy in Mice

Targeting FOLR1 with an approved antibody-drug conjugate restored anti-tumor immunity and enhanced anti-PD-1 effectiveness in colorectal cancer, overcoming immune evasion in mouse models.

Read at GEN (Genetic Engineering & Biotechnology News)

The Decoder Matthias Bastian SI

Google ships Gemini 4 Neon, its fourth new Gemini model in a month

Google released Gemini 4 Neon, a new frontier model built for software development, knowledge work, and science. It handles text, images, video, audio, and PDFs with a context window of one million tokens. At $3 per million input tokens and $15 per million output tokens, Neon is priced above previous Gemini 4 models. The article Google ships Gemini 4 Neon, its fourth new Gemini model in a month appeared first on The Decoder.

Read at The Decoder

arXiv (research papers) Upasana Dutta Tech policy

Asymmetric rhetorical strategies drive partisanship in television news

Television remains a dominant source of political information in the United States, yet systematic evidence on how partisanship is expressed in televised news remains limited. Prior research has largely treated partisanship as a one-dimensional construct, offering limited insight into the rhetorical strategies that underlie ideological bias. Here, by analyzing over 400,000 news episodes using large language models (LLMs), we provide a multi-dimensional measure of…

Read at arXiv (research papers)

AWS Machine Learning Blog Apoorva Chandra SI

How Salesforce is advancing global-scale model serving with cross-Region inference routing in Amazon SageMaker AI

Learn how Salesforce uses the cross-Region inference router in Amazon SageMaker AI to turn fragmented Regional capacity into reliable model serving. Salesforce adds capacity securely across Regions and accounts while keeping its data residency commitments, and sees per request where each call was served and why.

Read at AWS Machine Learning Blog

9to5Mac Ryan Christoffel Gadgets

New leaks detail next year’s iPhone Pro redesign and iPhone Duo 2

Next year’s fall iPhone lineup is expected to include 20th anniversary redesigned Pro models and an iPhone Duo 2. Here’s what to expect per the latest leaks.

Read at 9to5Mac

Older stories