Tootfinder

Alibaba Technical Report: Qwen3-VL beats GPT-5 and Gemini 2.5 Pro on visual tasks and has 100% accuracy on "needle-in-a-haystack" tests for 30-minute videos (Jonathan Kemper/The Decoder)
https://the-decoder.com/qwen3-vl-can-scan-two-hour-…

Qwen3-VL can scan two-hour videos and pinpoint nearly every detail
A few months after launching Qwen3-VL, Alibaba has released a detailed technical report on the open multimodal model. The data shows the system excels at image-based math tasks and can analyze hours of video footage.

#Anthropic Introduces #Skills for Custom Claude Tasks
https://www.infoq.com/news/2025/10/anthrop

VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications
Wei He, Yueqing Sun, Hongyan Hao, Xueyuan Hao, Zhikang Xia, Qi Gu, Chengcheng Han, Dengchang Zhao, Hui Su, Kefeng Zhang, Man Gao, Xi Su, Xiaodong Cai, Xunliang Cai, Yu Yang, Yunke Zhao
https://arxiv.org/abs/2509.26490

VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications
As LLM-based agents are increasingly deployed in real-life scenarios, existing benchmarks fail to capture their inherent complexity of handling extensive information, leveraging diverse resources, and managing dynamic user interactions. To address this gap, we introduce VitaBench, a challenging benchmark that evaluates agents on versatile interactive tasks grounded in real-world settings. Drawing from daily applications in food delivery, in-store consumption, and online travel services, VitaBen…

Personalized Auto-Grading and Feedback System for Constructive Geometry Tasks Using Large Language Models on an Online Math Platform
Yong Oh Lee, Byeonghun Bang, Joohyun Lee, Sejun Oh
https://arxiv.org/abs/2509.25529

Personalized Auto-Grading and Feedback System for Constructive Geometry Tasks Using Large Language Models on an Online Math Platform
As personalized learning gains increasing attention in mathematics education, there is a growing demand for intelligent systems that can assess complex student responses and provide individualized feedback in real time. In this study, we present a personalized auto-grading and feedback system for constructive geometry tasks, developed using large language models (LLMs) and deployed on the Algeomath platform, a Korean online tool designed for interactive geometric constructions. The proposed sys…

Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training
Junlin Han, Shengbang Tong, David Fan, Yufan Ren, Koustuv Sinha, Philip Torr, Filippos Kokkinos
https://arxiv.org/abs/2509.26625

Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training
Large Language Models (LLMs), despite being trained on text alone, surprisingly develop rich visual priors. These priors allow latent visual capabilities to be unlocked for vision tasks with a relatively small amount of multimodal data, and in some cases, to perform visual tasks without ever having seen an image. Through systematic analysis, we reveal that visual priors-the implicit, emergent knowledge about the visual world acquired during language pre-training-are composed of separable percep…

windsurfers: Windsurfers network (1986)
A network of interpersonal contacts among windsurfers in southern California during the Fall of 1986. The edge weights indicate the perception of social affiliations majored by the tasks in which each individual was asked​ to sort cards with other surfer’s name in the order of closeness.
This network has 43 nodes and 336 edges.
Tags: Social, Offline, Weighted

RANGER -- Repository-Level Agent for Graph-Enhanced Retrieval
Pratik Shah, Rajat Ghosh, Aryan Singhal, Debojyoti Dutta
https://arxiv.org/abs/2509.25257 https://

RANGER -- Repository-Level Agent for Graph-Enhanced Retrieval
General-purpose automated software engineering (ASE) includes tasks such as code completion, retrieval, repair, QA, and summarization. These tasks require a code retrieval system that can handle specific queries about code entities, or code entity queries (for example, locating a specific class or retrieving the dependencies of a function), as well as general queries without explicit code entities, or natural language queries (for example, describing a task and retrieving the corresponding code…

"As explained in chapter 11 of Meyer’s book, assertions are meant to check the correctness of a piece of software; that is, its ability to perform the tasks defined in their specification.
Because, you do have a specification, right? Right?"
https://deprogrammaticaipsum.com/asser…

Assertions
Colin Powell passed away during the preparation of this article. In one of the most memorable and abhorrent chapters of his career he gave a speech in the United Nations trying to justify the unjustifiable.

Artificial phantasia: Evidence for propositional reasoning-based mental imagery in large language models https://arxiv.org/abs/2509.23108 on the representation of visual imagery in humans; more information in the Bluesky thread

Artificial Phantasia: Evidence for Propositional Reasoning-Based Mental Imagery in Large Language Models
This study offers a novel approach for benchmarking complex cognitive behavior in artificial systems. Almost universally, Large Language Models (LLMs) perform best on tasks which may be included in their training data and can be accomplished solely using natural language, limiting our understanding of their emergent sophisticated cognitive capacities. In this work, we created dozens of novel items of a classic mental imagery task from cognitive psychology. A task which, traditionally, cognitive…

Caught a bug over the holidays so I’m mostly resting, feeling sorry for myself, and taking the time to at least carry out some mindless housekeeping tasks (updating dependencies, etc.) on some of my Node modules.
Released updates to the following packages yesterday:
Tape-based Node.js testing:
• Tap monkey (https://

tap-monkey
A tap formatter that’s also a monkey.

🎯 Real-world validation through extended #CCBench testing with human evaluators completing multi-turn tasks in isolated #Docker containers across frontend development, tool building, data analysis, testing & algorithms
🔧 Near parity with

It isn't all that hard for my reminders. I use the default stuff for now, but there are #opensource apps that can remind you to do things.

TimeScope: Towards Task-Oriented Temporal Grounding In Long Videos
Xiangrui Liu, Minghao Qin, Yan Shu, Zhengyang Liang, Yang Tian, Chen Jason Zhang, Bo Zhao, Zheng Liu
https://arxiv.org/abs/2509.26360 …

TimeScope: Towards Task-Oriented Temporal Grounding In Long Videos
Identifying key moments in long videos is essential for downstream understanding and reasoning tasks. In this paper, we introduce a new problem, Taskoriented Temporal Grounding ToTG, which aims to localize time intervals containing the necessary information based on a task's natural description. Along with the definition, we also present ToTG Bench, a comprehensive benchmark for evaluating the performance on ToTG. ToTG is particularly challenging for traditional approaches due to their limited …

Where to watch Texans vs. Broncos: TV channel, live stream, prediction, pick, odds, spread

https://www.cbssports.com/nfl/news/where-to-watch-texan…

The LLMs are useful for some tasks, I'm currently tidying up, proof reading and editing a document written by many international co-authors.
The LLM I'm using can very quickly correct grammar and spelling mistakes and produces much easier to understand text from occasionally tortured paragraphs. It still needs an experts eye (mine!) to check no mistakes have been introduced or complexities over-simplified.
This is actually the first time I've used an LLM for this task. It's making it much faster and less painful.
It also explains a lot about academic publishing lately..

SafeBehavior: Simulating Human-Like Multistage Reasoning to Mitigate Jailbreak Attacks in Large Language Models
Qinjian Zhao, Jiaqi Wang, Zhiqiang Gao, Zhihao Dou, Belal Abuhaija, Kaizhu Huang
https://arxiv.org/abs/2509.26345

SafeBehavior: Simulating Human-Like Multistage Reasoning to Mitigate Jailbreak Attacks in Large Language Models
Large Language Models (LLMs) have achieved impressive performance across diverse natural language processing tasks, but their growing power also amplifies potential risks such as jailbreak attacks that circumvent built-in safety mechanisms. Existing defenses including input paraphrasing, multi step evaluation, and safety expert models often suffer from high computational costs, limited generalization, or rigid workflows that fail to detect subtle malicious intent embedded in complex contexts. I…

Graphite: A GPU-Accelerated Mixed-Precision Graph Optimization Framework
Shishir Gopinath, Karthik Dantu, Steven Y. Ko
https://arxiv.org/abs/2509.26581 https://

Graphite: A GPU-Accelerated Mixed-Precision Graph Optimization Framework
We present Graphite, a GPU-accelerated nonlinear graph optimization framework. It provides a CUDA C++ interface to enable the sharing of code between a realtime application, such as a SLAM system, and its optimization tasks. The framework supports techniques to reduce memory usage, including in-place optimization, support for multiple floating point types and mixed-precision modes, and dynamically computed Jacobians. We evaluate Graphite on well-known bundle adjustment problems and find that it…

Pretrain-Test Task Alignment Governs Generalization in In-Context Learning
Mary I. Letey, Jacob A. Zavatone-Veth, Yue M. Lu, Cengiz Pehlevan
https://arxiv.org/abs/2509.26551 htt…

Pretrain-Test Task Alignment Governs Generalization in In-Context Learning
In-context learning (ICL) is a central capability of Transformer models, but the structures in data that enable its emergence and govern its robustness remain poorly understood. In this work, we study how the structure of pretraining tasks governs generalization in ICL. Using a solvable model for ICL of linear regression by linear attention, we derive an exact expression for ICL generalization error in high dimensions under arbitrary pretraining-testing task covariance mismatch. This leads to a…

"I'm an average user, so I don't need all the options and apps the programme has to offer. But, to be honest, #Microsoft is making it increasingly attractive to switch. Now that the company is putting #AI in everything, everything is becoming more annoying to use."
Can Dutch

The Invisible Mentor: Inferring User Actions from Screen Recordings to Recommend Better Workflows
Litao Yan, Andrew Head, Ken Milne, Vu Le, Sumit Gulwani, Chris Parnin, Emerson Murphy-Hill
https://arxiv.org/abs/2509.26557

The Invisible Mentor: Inferring User Actions from Screen Recordings to Recommend Better Workflows
Many users struggle to notice when a more efficient workflow exists in feature-rich tools like Excel. Existing AI assistants offer help only after users describe their goals or problems, which can be effortful and imprecise. We present InvisibleMentor, a system that turns screen recordings of task completion into vision-grounded reflections on tasks. It detects issues such as repetitive edits and recommends more efficient alternatives based on observed behavior. Unlike prior systems that rely o…

Enabling Time-Aware Priority Traffic Management over Distributed FPGA Nodes
Alberto Scionti, Paolo Savio, Francesco Lubrano, Federico Stirano, Antonino Nespola, Olivier Terzo, Corrado De Sio, Luca Sterpone
https://arxiv.org/abs/2509.26043

Enabling Time-Aware Priority Traffic Management over Distributed FPGA Nodes
Network Interface Cards (NICs) greatly evolved from simple basic devices moving traffic in and out of the network to complex heterogeneous systems offloading host CPUs from performing complex tasks on in-transit packets. These latter comprise different types of devices, ranging from NICs accelerating fixed specific functions (e.g., on-the-fly data compression/decompression, checksum computation, data encryption, etc.) to complex Systems-on-Chip (SoC) equipped with both general purpose processor…

RAE: A Neural Network Dimensionality Reduction Method for Nearest Neighbors Preservation in Vector Search
Han Zhang, Dongfang Zhao
https://arxiv.org/abs/2509.25839 https://

RAE: A Neural Network Dimensionality Reduction Method for Nearest Neighbors Preservation in Vector Search
While high-dimensional embedding vectors are being increasingly employed in various tasks like Retrieval-Augmented Generation and Recommendation Systems, popular dimensionality reduction (DR) methods such as PCA and UMAP have rarely been adopted for accelerating the retrieval process due to their inability of preserving the nearest neighbor (NN) relationship among vectors. Empowered by neural networks' optimization capability and the bounding effect of Rayleigh quotient, we propose a Regularize…

VoiceBridge: Designing Latent Bridge Models for General Speech Restoration at Scale
Chi Zhang, Zehua Chen, Kaiwen Zheng, Jun Zhu
https://arxiv.org/abs/2509.25275 https://…

VoiceBridge: Designing Latent Bridge Models for General Speech Restoration at Scale
Bridge models have recently been explored for speech enhancement tasks such as denoising, dereverberation, and super-resolution, while these efforts are typically confined to a single task or small-scale datasets, with constrained general speech restoration (GSR) capability at scale. In this work, we introduce VoiceBridge, a GSR system rooted in latent bridge models (LBMs), capable of reconstructing high-fidelity speech at full-band (\textit{i.e.,} 48~kHz) from various distortions. By compressi…

PAT: Pattern-Perceptive Transformer for Error Detection in Relational Databases
Jian Fu, Xixian Han, Xiaolong Wan, Wenjian Wang
https://arxiv.org/abs/2509.25907 https://

PAT: Pattern-Perceptive Transformer for Error Detection in Relational Databases
Error detection in relational databases is critical for maintaining data quality and is fundamental to tasks such as data cleaning and assessment. Current error detection studies mostly employ the multi-detector approach to handle heterogeneous attributes in databases, incurring high costs. Additionally, their data preprocessing strategies fail to leverage the variable-length characteristic of data sequences, resulting in reduced accuracy. In this paper, we propose an attribute-wise PAttern-per…

Evaluating the Impact of Radiographic Noise on Chest X-ray Semantic Segmentation and Disease Classification Using a Scalable Noise Injection Framework
Derek Jiu, Kiran Nijjer, Nishant Chinta, Ryan Bui, Ben Liu, Kevin Zhu
https://arxiv.org/abs/2509.25265

Evaluating the Impact of Radiographic Noise on Chest X-ray Semantic Segmentation and Disease Classification Using a Scalable Noise Injection Framework
Deep learning models are increasingly used for radiographic analysis, but their reliability is challenged by the stochastic noise inherent in clinical imaging. A systematic, cross-task understanding of how different noise types impact these models is lacking. Here, we evaluate the robustness of state-of-the-art convolutional neural networks (CNNs) to simulated quantum (Poisson) and electronic (Gaussian) noise in two key chest X-ray tasks: semantic segmentation and pulmonary disease classificati…

A human EEG dataset for multisensory perception and mental imagery #YOTO (You Only Think Once)

A Human EEG Dataset for Multisensory Perception and Mental Imagery - Scientific Data
The YOTO (You Only Think Once) dataset presents a human electroencephalog- raphy (EEG) resource for exploring multisensory perception and mental imagery. The study enrolled 26 participants who performed tasks involving both unimodal and multimodal stimuli. Researchers collected high-resolution EEG signals at a 1000 Hz sampling rate to capture high-temporal-resolution neural activity related to internal mental representations. The protocol incorporated visual, auditory, and combined cues to in…

Can LLMs Write Mathematics Papers? A Case Study in Reservoir Computing
Allen G Hart
https://arxiv.org/abs/2509.26550 https://arxiv.org/pdf/2509.26550

Can LLMs Write Mathematics Papers? A Case Study in Reservoir Computing
As AI capabilities continue to grow exponentially on economically relevant human expert tasks, with task completion horizons doubling every 7 months according to the Model Evaluation and Threat Research (METR), we are interested in how this applies to the task of mathematics research. To explore this, we evaluated the capability of four frontier large language models (LLMs), ChatGPT 5, Claude 4.1 Opus, Gemini 2.5 Pro, and Grok 4, at the task of creating a mini-paper on reservoir computing. All …

Are you afraid of our new GenAI overlords taking over our jobs soon? According to a new benchmark, The Remote Labor Index by Scale AI and the Center for AI Safety (CAIS), there's no need to be. The best current models are able to solve around ~2% of the tasks of the index: #AIResearch #GenAI

Replaced article(s) found for math.AC. https://arxiv.org/list/math.AC/new
[1/1]:
- Composing Global Solutions to Reasoning Tasks via Algebraic Objects in Neural Nets
Yuandong Tian

Commutative algebra neural network reveals genetic origins of diseases
JunJie Wee, Faisal Suwayyid, Mushal Zia, Hongsong Feng, Yuta Hozumi, Guo-Wei Wei
https://arxiv.org/abs/2509.26566

Commutative algebra neural network reveals genetic origins of diseases
Genetic mutations can disrupt protein structure, stability, and solubility, contributing to a wide range of diseases. Existing predictive models often lack interpretability and fail to integrate physical and chemical interactions critical to molecular mechanisms. Moreover, current approaches treat disease association, stability changes, and solubility alterations as separate tasks, limiting model generalizability. In this study, we introduce a unified framework based on multiscale commutative a…

A Chaotic Dynamics Framework Inspired by Dorsal Stream for Event Signal Processing
Yu Chen, Jing Lian, Zhaofei Yu, Jizhao Liu, Jisheng Dang, Gang Wang
https://arxiv.org/abs/2509.26085

A Chaotic Dynamics Framework Inspired by Dorsal Stream for Event Signal Processing
Event cameras are bio-inspired vision sensor that encode visual information with high dynamic range, high temporal resolution, and low latency.Current state-of-the-art event stream processing methods rely on end-to-end deep learning techniques. However, these models are heavily dependent on data structures, limiting their stability and generalization capabilities across tasks, thereby hindering their deployment in real-world scenarios. To address this issue, we propose a chaotic dynamics event …

Introducing Large Language Models in the Design Flow of Time Sensitive Networking
Rubi Debnath, Luxi Zhao, Mohammadreza Barzegaran, Sebastian Steinhorst
https://arxiv.org/abs/2509.26368

Introducing Large Language Models in the Design Flow of Time Sensitive Networking
The growing demand for real-time, safety-critical systems has significantly increased both the adoption and complexity of Time Sensitive Networking (TSN). Configuring an optimized TSN network is highly challenging, requiring careful planning, design, verification, validation, and deployment. Large Language Models (LLMs) have recently demonstrated strong capabilities in solving complex tasks, positioning them as promising candidates for automating end-to-end TSN deployment, referred to as TSN or…

TrackCore-F: Deploying Transformer-Based Subatomic Particle Tracking on FPGAs
Arjan Blankestijn, Uraz Odyurt, Amirreza Yousefzadeh
https://arxiv.org/abs/2509.26335 https://

TrackCore-F: Deploying Transformer-Based Subatomic Particle Tracking on FPGAs
The Transformer Machine Learning (ML) architecture has been gaining considerable momentum in recent years. In particular, computational High-Energy Physics tasks such as jet tagging and particle track reconstruction (tracking), have either achieved proper solutions, or reached considerable milestones using Transformers. On the other hand, the use of specialised hardware accelerators, especially FPGAs, is an effective method to achieve online, or pseudo-online latencies. The development and inte…

🔮 The Future: Z1 Scale-Up Coming next: Z1 chips with 250K interconnected Pits per chip. Powers energy-based models 10,000x more efficient than #GPUs for generative tasks. Check the new paper on Denoising Thermodynamic #Models!
📺 Watch the full video:

PRISM: Progressive Rain removal with Integrated State-space Modeling
Pengze Xue, Shanwen Wang, Fei Zhou, Yan Cui, Xin Sun
https://arxiv.org/abs/2509.26413 https://

PRISM: Progressive Rain removal with Integrated State-space Modeling
Image deraining is an essential vision technique that removes rain streaks and water droplets, enhancing clarity for critical vision tasks like autonomous driving. However, current single-scale models struggle with fine-grained recovery and global consistency. To address this challenge, we propose Progressive Rain removal with Integrated State-space Modeling (PRISM), a progressive three-stage framework: Coarse Extraction Network (CENet), Frequency Fusion Network (SFNet), and Refine Network (RNe…

Kiedy spałeś jeszcze krócej niż zwykle, bo tuż przed snem wpadłeś na to, jak zaimplementować mnożenie w https://nandgame.com.
Oczywiście, jak już się położyłeś, to wpadłeś na to, jak zrobić to szybciej, i z 32-bitowym wynikiem. Aczkolwiek stara metoda też dobra, taka ładnie pedagogiczna.

STaR-Attack: A Spatio-Temporal and Narrative Reasoning Attack Framework for Unified Multimodal Understanding and Generation Models
Shaoxiong Guo, Tianyi Du, Lijun Li, Yuyao Wu, Jie Li, Jing Shao
https://arxiv.org/abs/2509.26473

STaR-Attack: A Spatio-Temporal and Narrative Reasoning Attack Framework for Unified Multimodal Understanding and Generation Models
Unified Multimodal understanding and generation Models (UMMs) have demonstrated remarkable capabilities in both understanding and generation tasks. However, we identify a vulnerability arising from the generation-understanding coupling in UMMs. The attackers can use the generative function to craft an information-rich adversarial image and then leverage the understanding function to absorb it in a single pass, which we call Cross-Modal Generative Injection (CMGI). Current attack methods on mali…

Fast-dLLM v2: Efficient Block-Diffusion LLM
Chengyue Wu, Hao Zhang, Shuchen Xue, Shizhe Diao, Yonggan Fu, Zhijian Liu, Pavlo Molchanov, Ping Luo, Song Han, Enze Xie
https://arxiv.org/abs/2509.26328

Fast-dLLM v2: Efficient Block-Diffusion LLM
Autoregressive (AR) large language models (LLMs) have achieved remarkable performance across a wide range of natural language tasks, yet their inherent sequential decoding limits inference efficiency. In this work, we propose Fast-dLLM v2, a carefully designed block diffusion language model (dLLM) that efficiently adapts pretrained AR models into dLLMs for parallel text generation, requiring only approximately 1B tokens of fine-tuning. This represents a 500x reduction in training data compared …

fev-bench: A Realistic Benchmark for Time Series Forecasting
Oleksandr Shchur, Abdul Fatir Ansari, Caner Turkmen, Lorenzo Stella, Nick Erickson, Pablo Guerron, Michael Bohlke-Schneider, Yuyang Wang
https://arxiv.org/abs/2509.26468

fev-bench: A Realistic Benchmark for Time Series Forecasting
Benchmark quality is critical for meaningful evaluation and sustained progress in time series forecasting, particularly given the recent rise of pretrained models. Existing benchmarks often have narrow domain coverage or overlook important real-world settings, such as tasks with covariates. Additionally, their aggregation procedures often lack statistical rigor, making it unclear whether observed performance differences reflect true improvements or random variation. Many benchmarks also fail to…

Are neural scaling laws leading quantum chemistry astray?
Siwoo Lee, Adji Bousso Dieng
https://arxiv.org/abs/2509.26397 https://arxiv.org/pdf/2509.26397

Are neural scaling laws leading quantum chemistry astray?
Neural scaling laws are driving the machine learning community toward training ever-larger foundation models across domains, assuring high accuracy and transferable representations for extrapolative tasks. We test this promise in quantum chemistry by scaling model capacity and training data from quantum chemical calculations. As a generalization task, we evaluate the resulting models' predictions of the bond dissociation energy of neutral H$_2$, the simplest possible molecule. We find that, reg…

Since the start of the semester I feel like being hunted by an angry pack of looming deadlines: everything seems to cluster around October 31 and mid-November.
But! I have now finished all work-related tasks due October 31, and it's only the 28!

Since the start of the semester I feel like being hunted by an angry pack of looming deadlines: everything seems to cluster around October 31 and mid-November.
But! I have now finished all work-related tasks due October 31, and it's only the 28!

Since the start of the semester I feel like being hunted by an angry pack of looming deadlines: everything seems to cluster around October 31 and mid-November.
But! I have now finished all work-related tasks due October 31, and it's only the 28!

windsurfers: Windsurfers network (1986)
A network of interpersonal contacts among windsurfers in southern California during the Fall of 1986. The edge weights indicate the perception of social affiliations majored by the tasks in which each individual was asked​ to sort cards with other surfer’s name in the order of closeness.
This network has 43 nodes and 336 edges.
Tags: Social, Offline, Weighted

Red Teaming Program Repair Agents: When Correct Patches can Hide Vulnerabilities
Simin Chen, Yixin He, Suman Jana, Baishakhi Ray
https://arxiv.org/abs/2509.25894 https://…

Red Teaming Program Repair Agents: When Correct Patches can Hide Vulnerabilities
LLM-based agents are increasingly deployed for software maintenance tasks such as automated program repair (APR). APR agents automatically fetch GitHub issues and use backend LLMs to generate patches that fix the reported bugs. However, existing work primarily focuses on the functional correctness of APR-generated patches, whether they pass hidden or regression tests, while largely ignoring potential security risks. Given the openness of platforms like GitHub, where any user can raise issues an…

I so hope this blows up in his face spectacularly.
Note that I do not believe that any LLM can become as skilled as the least skilled #InfoSec professional using conventional tools, so I’m quite confident this will not reach any useful goal.
I’m saying he needs to learn a lesson, and it would be great if it were useful for others as well. @…

Auto-ARGUE: LLM-Based Report Generation Evaluation
William Walden, Marc Mason, Orion Weller, Laura Dietz, Hannah Recknor, Bryan Li, Gabrielle Kaili-May Liu, Yu Hou, James Mayfield, Eugene Yang
https://arxiv.org/abs/2509.26184

Auto-ARGUE: LLM-Based Report Generation Evaluation
Generation of long-form, citation-backed reports is a primary use case for retrieval augmented generation (RAG) systems. While open-source evaluation tools exist for various RAG tasks, ones tailored to report generation are lacking. Accordingly, we introduce Auto-ARGUE, a robust LLM-based implementation of the recent ARGUE framework for report generation evaluation. We present analysis of Auto-ARGUE on the report generation pilot task from the TREC 2024 NeuCLIR track, showing good system-level …

Today I'm heading to #Berlin again with tiny sh0rky 🦈🤏 and some cleanup tasks left for the ride.
The intel_fw library turns out to be very rich and powerful already.
I will start my new job on Monday, and I'll be using #Arch btw.

AGOCS -- Accurate Google Cloud Simulator Framework
Leszek Sliwko, Vladimir Getov
https://arxiv.org/abs/2509.26120 https://arxiv.org/pdf/2509.26120

AGOCS -- Accurate Google Cloud Simulator Framework
This paper presents the Accurate Google Cloud Simulator (AGOCS) - a novel high-fidelity Cloud workload simulator based on parsing real workload traces, which can be conveniently used on a desktop machine for day-to-day research. Our simulation is based on real-world workload traces from a Google Cluster with 12.5K nodes, over a period of a calendar month. The framework is able to reveal very precise and detailed parameters of the executed jobs, tasks and nodes as well as to provide actual resou…

scUnified: An AI-Ready Standardized Resource for Single-Cell RNA Sequencing Analysis
Ping Xu, Zaitian Wang, Zhirui Wang, Pengjiang Li, Ran Zhang, Gaoyang Li, Hanyu Xie, Jiajia Wang, Yuanchun Zhou, Pengfei Wang
https://arxiv.org/abs/2509.25884

scUnified: An AI-Ready Standardized Resource for Single-Cell RNA Sequencing Analysis
Single-cell RNA sequencing (scRNA-seq) technology enables systematic delineation of cellular states and interactions, providing crucial insights into cellular heterogeneity. Building on this potential, numerous computational methods have been developed for tasks such as cell clustering, cell type annotation, and marker gene identification. To fully assess and compare these methods, standardized, analysis-ready datasets are essential. However, such datasets remain scarce, and variations in data …

Welcome to the world of the field, engineering.
For a long time, we've hired very few into sales, mktg, support or consulting that don't already gobs of experience elsewhere.
✅ How #Microsoft’s developers are using #AI - The Verge

How Microsoft’s developers are using AI
Microsoft is pushing its software developers to use AI internally. It’s part of an effort to get more AI agents doing the boring developer tasks.

DEPTHOR : Robust Depth Enhancement from a Real-World Lightweight dToF and RGB Guidance
Jijun Xiang, Longliang Liu, Xuan Zhu, Xianqi Wang, Min Lin, Xin Yang
https://arxiv.org/abs/2509.26498

DEPTHOR++: Robust Depth Enhancement from a Real-World Lightweight dToF and RGB Guidance
Depth enhancement, which converts raw dToF signals into dense depth maps using RGB guidance, is crucial for improving depth perception in high-precision tasks such as 3D reconstruction and SLAM. However, existing methods often assume ideal dToF inputs and perfect dToF-RGB alignment, overlooking calibration errors and anomalies, thus limiting real-world applicability. This work systematically analyzes the noise characteristics of real-world lightweight dToF sensors and proposes a practical and n…

From NL2SQL to NL2GeoSQL: GeoSQL-Eval for automated evaluation of LLMs on PostGIS queries
Shuyang Hou, Haoyue Jiao, Ziqi Liu, Lutong Xie, Guanyu Chen, Shaowen Wu, Xuefeng Guan, Huayi Wu
https://arxiv.org/abs/2509.25264

From NL2SQL to NL2GeoSQL: GeoSQL-Eval for automated evaluation of LLMs on PostGIS queries
In recent years, large language models (LLMs) have achieved remarkable progress in natural language understanding and structured query generation (NL2SQL). However, extending these advances to GeoSQL tasks in the PostGIS environment remains challenging due to the complexity of spatial functions, geometric data types, and execution semantics. Existing evaluations primarily focus on general relational databases or Google Earth Engine code generation, leaving a lack of systematic benchmarks tailor…

OpenAI releases gpt-oss-safeguard, its open-weight reasoning models for safety classification tasks, available in 120B and 20B parameters, under Apache 2.0 (OpenAI)
https://openai.com/index/introducing-gpt-oss-safeguard/

Diversity-Incentivized Exploration for Versatile Reasoning
Zican Hu, Shilin Zhang, Yafu Li, Jianhao Yan, Xuyang Hu, Leyang Cui, Xiaoye Qu, Chunlin Chen, Yu Cheng, Zhi Wang
https://arxiv.org/abs/2509.26209

Diversity-Incentivized Exploration for Versatile Reasoning
Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a crucial paradigm for incentivizing reasoning capabilities in Large Language Models (LLMs). Due to vast state-action spaces and reward sparsity in reasoning tasks, existing methods often struggle with deficient exploration and poor sample efficiency. In the paper, we propose \textbf{DIVER} (\textbf{D}iversity-\textbf{I}ncentivized Exploration for \textbf{V}ersatil\textbf{E} \textbf{R}easoning), an innovative framework that hi…

Crosslisted article(s) found for cs.LG. https://arxiv.org/list/cs.LG/new
[3/7]:
- Personalized Auto-Grading and Feedback System for Constructive Geometry Tasks Using Large Languag...
Yong Oh Lee, Byeonghun Bang, Joohyun Lee, Sejun Oh

Towards Reliable Benchmarking: A Contamination Free, Controllable Evaluation Framework for Multi-step LLM Function Calling
Seiji Maekawa, Jackson Hassell, Pouya Pezeshkpour, Tom Mitchell, Estevam Hruschka
https://arxiv.org/abs/2509.26553

Towards Reliable Benchmarking: A Contamination Free, Controllable Evaluation Framework for Multi-step LLM Function Calling
As language models gain access to external tools via structured function calls, they become increasingly more capable of solving complex, multi-step tasks. However, existing benchmarks for tool-augmented language models (TaLMs) provide insufficient control over factors such as the number of functions accessible, task complexity, and input size, and remain vulnerable to data contamination. We present FuncBenchGen, a unified, contamination-free framework that evaluates TaLMs by generating synthet…

Protocode: Prototype-Driven Interpretability for Code Generation in LLMs
Krishna Vamshi Bodla, Haizhao Yang
https://arxiv.org/abs/2509.25247 https://arxiv.…

Protocode: Prototype-Driven Interpretability for Code Generation in LLMs
Since the introduction of Large Language Models (LLMs), they have been widely adopted for various tasks such as text summarization, question answering, speech-to-text translation, and more. In recent times, the use of LLMs for code generation has gained significant attention, with tools such as Cursor and Windsurf demonstrating the ability to analyze massive code repositories and recommend relevant changes. Big tech companies have also acknowledged the growing reliance on LLMs for code generati…

RE: https://hachyderm.io/@thomasfuchs/115601979925351548
hear me out, how about NOT CRAMMING EVERYTHING INTO ONE DEVICE that just works mid for everything, but instead, you know, do some actual innovation here and there
for example, make devices specifically tailored for certain tasks
like if you're Apple why in the fuck don't you make devices with e-paper screens for people who don't want to be terminally online

Another day another study showing that "AI assistants" do not work reliably for cognitive tasks.
"Largest study of its kind shows AI assistants misrepresent news content 45% of the time – regardless of language or territory"
https://www.bbc.co.uk/mediacentre/202…

Largest study of its kind shows AI assistants misrepresent news content 45% of the time – regardless of language or territory
An intensive international study was coordinated by the European Broadcasting Union (EBU) and led by the BBC

from my link log —
Two workers are quadratically better than one: modelling a queue with PRISM.
https://www.hillelwayne.com/post/queueing-prism/
saved 2020-11-02

Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark
Minhui Zhu, Minyang Tian, Xiaocheng Yang, Tianci Zhou, Penghao Zhu, Eli Chertkov, Shengyan Liu, Yufeng Du, Lifan Yuan, Ziming Ji, Indranil Das, Junyi Cao, Yufeng Du, Jinchen He, Yifan Su, Jiabin Yu, Yikun Jiang, Yujie Zhang, Chang Liu, Ze-Min Huang, Weizhen Jia, Xinan Chen, Peixue Wu, Yunkai Wang, Juntai Zhou, Yong Zhao, Farshid Jafarpour, Jessie Shelton, Aaron Young, John Bartolotta, Wenchao Xu,…

Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark
While large language models (LLMs) with reasoning capabilities are progressing rapidly on high-school math competitions and coding, can they reason effectively through complex, open-ended challenges found in frontier physics research? And crucially, what kinds of reasoning tasks do physicists want LLMs to assist with? To address these questions, we present the CritPt (Complex Research using Integrated Thinking - Physics Test, pronounced "critical point"), the first benchmark designed to test LL…

Like all the rest of the nerds, I did a bit of tech support on family computers.
They're all popping up windows from scam virus scanners lying that subscriptions need to be renewed or machines are unprotected. People don't know how to remove these things. Luckily they also don't really know how to pay the subscription.
Their phones are updating on them. Changing where buttons used to be. Removing options. Forcing people to register to use they things they have been doing for years.
They don't know how to register.
Things pop up asking for passwords and they have no idea who is asking or which password to use.
I tell them that I don't really understand why they keep using Windows now it is so shitty and awful. They say they don't know how to use anything else. The fact they don't really know how to use windows either doesn't seem to register.
The tech corporations have given up completely on being user friendly. They are all deliberately user hostile and exploitative now.
Corporate tech is terrible. The industry is failing it's users, abusing them. People don't even know there is any other way. They are just giving up on achieving their tasks until someone can fix the pop-ups and subscription boxes and passwords and 2fa for them.
Tech sucks now. Sucks hard.
#tech #christmasTechSupport

Been using a number of AI models over the past week or so as work has slowed down, giving me time to explore things more deeply.
Been using Claude Code with musistudio/claude-code-router which is great as I can switch between different models on similar tasks.
Experience so far has been that Gemini 3 Flash is very good for thinking and coding tasks but the code does tend to be fragile so rewrites are needed. For tough problems where the errors are not straightforward it falls d…

VisualOverload: Probing Visual Understanding of VLMs in Really Dense Scenes
Paul Gavrikov, Wei Lin, M. Jehanzeb Mirza, Soumya Jahagirdar, Muhammad Huzaifa, Sivan Doveh, Serena Yeung-Levy, James Glass, Hilde Kuehne
https://arxiv.org/abs/2509.25339

VisualOverload: Probing Visual Understanding of VLMs in Really Dense Scenes
Is basic visual understanding really solved in state-of-the-art VLMs? We present VisualOverload, a slightly different visual question answering (VQA) benchmark comprising 2,720 question-answer pairs, with privately held ground-truth responses. Unlike prior VQA datasets that typically focus on near global image understanding, VisualOverload challenges models to perform simple, knowledge-free vision tasks in densely populated (or, overloaded) scenes. Our dataset consists of high-resolution scans …

Replaced article(s) found for cs.CL. https://arxiv.org/list/cs.CL/new
[4/5]:
- Composing Global Solutions to Reasoning Tasks via Algebraic Objects in Neural Nets
Yuandong Tian

'Cost-Effective Machine Learning for Automatically Processing Bibliographic Metadata' is a very readable account of using DistilBERT for specific DH tasks https://www.euppublishing.com/doi/full/10.3366/ijhac.2025.0353

Reinforcement Learning-Guided Chain-of-Draft for Token-Efficient Code Generation
Xunzhu Tang, Iyiola Emmanuel Olatunji, Tiezhu Sun, Jacques Klein, Tegawende F. Bissyande
https://arxiv.org/abs/2509.25243

Reinforcement Learning-Guided Chain-of-Draft for Token-Efficient Code Generation
LLMs demonstrate surface-level fluency in code generation but struggle with structured reasoning tasks requiring correctness and semantic alignment. While Chain-of-Thought (CoT) prompting enhances reasoning through intermediate steps, it suffers from verbosity and inefficiency. Chain-of-Draft (CoD) prompting offers more concise reasoning, but the stochastic nature of LLMs produces varying solution quality, making optimal selection challenging. We propose \multicod, a reinforcement learning fram…

Todoist FTW.
Several years ago we had smoky pies and rolls for Thanksgiving because we put off cleaning the ovens until it was too late.
I created a recurring Todoist task to clean the ovens on the 3rd Sunday of November. Ever since then we’ve had pristine ovens and smoke free cooking every Thanksgiving.
Collette appreciates that a holiday for which most of the tasks are hers, she doesn’t have to worry and the ovens are ready to go.

When you develop a hippier mindset, even the smallest daily tasks start taking hours longer, because in the end, none of it really matters anyway, I mean, why rush?!
#MentalHealth #Hippie

SCUBA: Salesforce Computer Use Benchmark
Yutong Dai, Krithika Ramakrishnan, Jing Gu, Matthew Fernandez, Yanqi Luo, Viraj Prabhu, Zhenyu Hu, Silvio Savarese, Caiming Xiong, Zeyuan Chen, Ran Xu
https://arxiv.org/abs/2509.26506

SCUBA: Salesforce Computer Use Benchmark
We introduce SCUBA, a benchmark designed to evaluate computer-use agents on customer relationship management (CRM) workflows within the Salesforce platform. SCUBA contains 300 task instances derived from real user interviews, spanning three primary personas, platform administrators, sales representatives, and service agents. The tasks test a range of enterprise-critical abilities, including Enterprise Software UI navigation, data manipulation, workflow automation, information retrieval, and tro…

I am trying to alternate between using Workflowy and Emacs Org for tasks. The tool chosen randomly on a weekly basis.
(Intention is to understand what features I find absolutely required).
#Emacs #orgmode #workflowy

You know that you’re teaching in #Switzerland when you catch the majority of students running the live broadcast of today’s #ski competition in a separate window while doing learning tasks
#AcademicChatter

Deconstructing Self-Bias in LLM-generated Translation Benchmarks
Wenda Xu, Sweta Agrawal, Vil\'em Zouhar, Markus Freitag, Daniel Deutsch
https://arxiv.org/abs/2509.26600 htt…

Deconstructing Self-Bias in LLM-generated Translation Benchmarks
As large language models (LLMs) begin to saturate existing benchmarks, automated benchmark creation using LLMs (LLM as a benchmark) has emerged as a scalable alternative to slow and costly human curation. While these generated test sets have to potential to cheaply rank models, we demonstrate a critical flaw. LLM generated benchmarks systematically favor the model that created the benchmark, they exhibit self bias on low resource languages to English translation tasks. We show three key finding…

A Multi-Language Object-Oriented Programming Benchmark for Large Language Models
Shuai Wang, Liang Ding, Li Shen, Yong Luo, Han Hu, Lefei Zhang, Fu Lin
https://arxiv.org/abs/2509.26111

A Multi-Language Object-Oriented Programming Benchmark for Large Language Models
Establishing fair and robust benchmarks is essential for evaluating intelligent code generation by large language models (LLMs). Our survey of 35 existing benchmarks uncovers three major imbalances: 85.7% focus on a single programming language; 94.3% target only function-level or statement-level tasks; and over 80% include fewer than ten test cases on average. To address these gaps, we propose MultiOOP, a multi-language object-oriented programming benchmark covering six popular languages (Pytho…

AI Playing Business Games: Benchmarking Large Language Models on Managerial Decision-Making in Dynamic Simulations
Berdymyrat Ovezmyradov
https://arxiv.org/abs/2509.26331 https:…

AI Playing Business Games: Benchmarking Large Language Models on Managerial Decision-Making in Dynamic Simulations
The rapid advancement of LLMs sparked significant interest in their potential to augment or automate managerial functions. One of the most recent trends in AI benchmarking is performance of Large Language Models (LLMs) over longer time horizons. While LLMs excel at tasks involving natural language and pattern recognition, their capabilities in multi-step, strategic business decision-making remain largely unexplored. Few studies demonstrated how results can be different from benchmarks in short-…

QUARTZ : QA-based Unsupervised Abstractive Refinement for Task-oriented Dialogue Summarization
Mohamed Imed Eddine Ghebriout (Universite de Lorraine, CNRS, Inria, LORIA, Nancy, France), Ga\"el Guibon (Universite de Lorraine, CNRS, Inria, LORIA, Nancy, France), Ivan Lerner (Inserm, Centre de Recherche des Cordeliers, Universite Paris Cite, Sorbonne Universite, Paris, France), Emmanuel Vincent (Universite de Lorraine, CNRS, Inria, LORIA, Nancy, France)

QUARTZ : QA-based Unsupervised Abstractive Refinement for Task-oriented Dialogue Summarization
Dialogue summarization aims to distill the core meaning of a conversation into a concise text. This is crucial for reducing the complexity and noise inherent in dialogue-heavy applications. While recent approaches typically train language models to mimic human-written summaries, such supervision is costly and often results in outputs that lack task-specific focus limiting their effectiveness in downstream applications, such as medical tasks. In this paper, we propose \app, a framework for task-…

Some people now argue that LLMs are useless. I disagree; they can be very useful if you take them as what they are: models of language that generate text on the basis of some given text. As such, they can be useful for a wide range of text-related tasks, including assisting with writing. And the more formulaic the genre, the better they work obviously. This is part of the reason why they are so popular with students, and in academia more generally.
⇢

Some people now argue that LLMs are useless. I disagree; they can be very useful if you take them as what they are: models of language that generate text on the basis of some given text. As such, they can be useful for a wide range of text-related tasks, including assisting with writing. And the more formulaic the genre, the better they work obviously. This is part of the reason why they are so popular with students, and in academia more generally.
⇢

Automatically Generating Web Applications from Requirements Via Multi-Agent Test-Driven Development
Yuxuan Wan, Tingshuo Liang, Jiakai Xu, Jingyu Xiao, Yintong Huo, Michael R. Lyu
https://arxiv.org/abs/2509.25297

Automatically Generating Web Applications from Requirements Via Multi-Agent Test-Driven Development
Developing full-stack web applications is complex and time-intensive, demanding proficiency across diverse technologies and frameworks. Although recent advances in multimodal large language models (MLLMs) enable automated webpage generation from visual inputs, current solutions remain limited to front-end tasks and fail to deliver fully functional applications. In this work, we introduce TDDev, the first test-driven development (TDD)-enabled LLM-agent framework for end-to-end full-stack web app…

OntoAligner Meets Knowledge Graph Embedding Aligners
Hamed Babaei Giglou, Jennifer D'Souza, S\"oren Auer, Mahsa Sanaei
https://arxiv.org/abs/2509.26417 https://

OntoAligner Meets Knowledge Graph Embedding Aligners
Ontology Alignment (OA) is essential for enabling semantic interoperability across heterogeneous knowledge systems. While recent advances have focused on large language models (LLMs) for capturing contextual semantics, this work revisits the underexplored potential of Knowledge Graph Embedding (KGE) models, which offer scalable, structure-aware representations well-suited to ontology-based tasks. Despite their effectiveness in link prediction, KGE methods remain underutilized in OA, with most p…

Uber plans to launch data labelling tasks in the US for some drivers to earn extra money, appearing under "digital tasks" in the driver app, later this fall (Natalie Lung/Bloomberg)
https://www.bloomberg.com/news/articles/20

LLM Agents for Knowledge Discovery in Atomic Layer Processing
Andreas Werbrouck, Marshall B. Lindsay, Matthew Maschmann, Matthias J. Young
https://arxiv.org/abs/2509.26201 https…

LLM Agents for Knowledge Discovery in Atomic Layer Processing
Large Language Models (LLMs) have garnered significant attention for several years now. Recently, their use as independently reasoning agents has been proposed. In this work, we test the potential of such agents for knowledge discovery in materials science. We repurpose LangGraph's tool functionality to supply agents with a black box function to interrogate. In contrast to process optimization or performing specific, user-defined tasks, knowledge discovery consists of freely exploring the syste…

Towards Repository-Level Program Verification with Large Language Models
Si Cheng Zhong, Xujie Si
https://arxiv.org/abs/2509.25197 https://arxiv.org/pdf/25…

Towards Repository-Level Program Verification with Large Language Models
Recent advancements in large language models (LLMs) suggest great promises in code and proof generations. However, scaling automated formal verification to real-world projects requires resolving cross-module dependencies and global contexts, which are crucial challenges overlooked by existing LLM-based methods with a special focus on targeting isolated, function-level verification tasks. To systematically explore and address the significant challenges of verifying entire software repositories, …

Sunday Robotics unveils Memo, a fully autonomous home robot capable of tasks like making espresso and loading dishwashers, set to launch in beta in 2026 (Will Knight/Wired)
https://www.wired.com/story/memo-sunday-robotics-home-robot/

This Home Robot Clears Tables and Loads the Dishwasher All by Itself
Sunday Robotics has a new way to train robots to do common household tasks. The startup plans to put its fully autonomous robots in homes next year.

🚀 Real results: Geoffrey Huntley ran a 3-month loop building a complete programming language. YC hackathon teams shipped 6 repos overnight for $297 in API costs
✅ Best for: Large refactors, batch operations, test coverage, documentation generation - tasks with clear completion criteria
⚠️ Not for: Ambiguous requirements, architectural decisions, security-sensitive code, or exploration work

Creation, Critique, and Consumption: Exploring Generative AI Descriptions for Supporting Blind and Low Vision Professionals with Visual Tasks
Lucy Jiang, Lotus Zhang, Leah Findlater
https://arxiv.org/abs/2510.08991

Creation, Critique, and Consumption: Exploring Generative AI Descriptions for Supporting Blind and Low Vision Professionals with Visual Tasks
Many blind and low vision (BLV) people are excluded from professional roles that may involve visual tasks due to access barriers and persisting stigmas. Advancing generative AI systems can support BLV people through providing contextual and personalized visual descriptions for creation, critique, and consumption. In this workshop paper, we provide design suggestions for how visual descriptions can be better contextualized for multiple professional tasks. We conclude by discussing how these desi…

windsurfers: Windsurfers network (1986)
A network of interpersonal contacts among windsurfers in southern California during the Fall of 1986. The edge weights indicate the perception of social affiliations majored by the tasks in which each individual was asked​ to sort cards with other surfer’s name in the order of closeness.
This network has 43 nodes and 336 edges.
Tags: Social, Offline, Weighted

Arbiter, which is using AI to automate healthcare administrative tasks, emerges from stealth with a $52M seed from multiple family offices at a $400M valuation (Rebecca Torrence/Business Insider)
https://www.businessinsider.com/health-sta

This former Eli Lilly exec just raised a $52 million seed round from family offices to build a new healthcare AI startup.
Thirty Madison's former president is launching her own healthcare AI startup without a drop of VC funding.

Benchmark It Yourself (BIY): Preparing a Dataset and Benchmarking AI Models for Scatterplot-Related Tasks
Jo\~ao Palmeiro, Diogo Duarte, Rita Costa, Pedro Bizarro
https://arxiv.org/abs/2510.06071

Benchmark It Yourself (BIY): Preparing a Dataset and Benchmarking AI Models for Scatterplot-Related Tasks
AI models are increasingly used for data analysis and visualization, yet benchmarks rarely address scatterplot-specific tasks, limiting insight into performance. To address this gap for one of the most common chart types, we introduce a synthetic, annotated dataset of over 18,000 scatterplots from six data generators and 17 chart designs, and a benchmark based on it. We evaluate proprietary models from OpenAI and Google using N-shot prompting on five distinct tasks derived from annotations of c…

Beijing-based DP Technology, which develops AI tools used by researchers for tasks like computer-aided drug design and battery design, raised a ~$114M Series C (Eunice Xu/South China Morning Post)
https://www.scmp.com/business/companies/ar

AI-for-Science start-up DP Technology raises US$114 million in Series C round
The Beijing-based AI-for-Science firm said the Series C round would fund hiring and R&D, as interest grows in using AI to speed up scientific discovery.

Anthropic finds that LLMs trained to "reward hack" by cheating on coding tasks show even more misaligned behavior, including sabotaging AI-safety research (Anthropic)
https://www.anthropic.com/research/emergent-misalignment-reward-hacking

From shortcuts to sabotage: natural emergent misalignment from reward hacking
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

China's MiniMax releases M2.1, an upgrade to its open-source M2 model that it says has "significantly enhanced" coding capabilities in Rust, Java, and others (MiniMax)
https://www.minimax.io/news/minimax-m21

MiniMax M2.1: Significantly Enhanced Multi-Language Programming, Built for Real-World Complex Tasks - MiniMax News

Evaluating the Quality of Randomness and Entropy in Tasks Supported by Large Language Models
Rabimba Karanjai, Yang Lu, Ranjith Chodavarapu, Lei Xu, Weidong Shi
https://arxiv.org/abs/2510.12080

Evaluating the Quality of Randomness and Entropy in Tasks Supported by Large Language Models
The rapid advancement of large language model (LLM) technology has led to diverse applications, many of which inherently require randomness, such as stochastic decision-making, gaming, scheduling, AI agents, and cryptography-related tasks. However, the capabilities of LLMs in handling randomness, particularly in generating and utilizing random numbers effectively, remain unclear. This paper investigates the capacity of LLMs for handling tasks that involve randomness through a series of experime…

Selection, Reflection and Self-Refinement: Revisit Reasoning Tasks via a Causal Lens
Yunlong Deng, Boyang Sun, Yan Li, Lingjing Kong, Zeyu Tang, Kun Zhang, Guangyi Chen
https://arxiv.org/abs/2510.08222

Selection, Reflection and Self-Refinement: Revisit Reasoning Tasks via a Causal Lens
Due to their inherent complexity, reasoning tasks have long been regarded as rigorous benchmarks for assessing the capabilities of machine learning models, especially large language models (LLMs). Although humans can solve these tasks with ease, existing models, even after extensive pre-training and post-training at scale, still fail to perform reasoning reliably. In this paper, we revisit reasoning tasks from a causal perspective, seeking to understand their behavior in latent space and to off…

ChatGPT Atlas hands-on: generally able to interpret instructions and navigate simple menus, but "technical constraints on session length" are a limiting factor (Kyle Orland/Ars Technica)
https://arstechnica.com/features/2025/

We let OpenAI’s “Agent Mode” surf the web for us—here’s what happened
From scanning emails to building fansites, Atlas can ably automate some web-based tasks.

Hyro, whose AI agents let US health care organizations automate tasks like scheduling and prescription renewals, raised $45M, taking its total funding to $95M (Sophie Shulman/CTech)
https://www.calcalistech.com/ctechnews/article/rylcu6valg

Hyro secures $45 million to scale AI automation in US healthcare
The startup aims to ease staffing pressures and improve patient access by deploying AI agents across hospitals and health systems nationwide.

The Pentagon partners with xAI to embed the company's frontier AI systems, based on the Grok family of models, directly into GenAI.mil as soon as early 2026 (Bonny Chu/Fox News)
https://www.foxnews.com/politics/pentagon-

Pentagon taps Musk's xAI to boost sensitive government workflows, support military operations
Pentagon partners with Elon Musk's xAI to deploy Grok AI across military systems, giving 3 million personnel advanced AI tools by 2026 for sensitive tasks.

Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks
Yuxiang Zhang, Jiangming Shu, Ye Ma, Xueyuan Lin, Shangxi Wu, Jitao Sang
https://arxiv.org/abs/2510.12635

Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks
Large Language Models face challenges in long-horizon agentic tasks as their constrained memory is easily overwhelmed by distracting or irrelevant context. Existing working memory methods typically rely on external, heuristic mechanisms that are decoupled from the agent's core policy. In this work, we reframe working memory management as a learnable, intrinsic capability. We propose a novel framework, Memory-as-Action, where an agent actively manages its working memory by executing explicit edi…

METR: Claude Opus 4.5 has a 50% task completion time horizon of about 4 hours and 49 minutes, more than double that of Claude Opus 4 released earlier this year (@metr_evals)
https://x.com/metr_evals/status/2002203627377574113

METR (@METR_Evals) on X
We estimate that, on our tasks, Claude Opus 4.5 has a 50%-time horizon of around 4 hrs 49 mins (95% confidence interval of 1 hr 49 mins to 20 hrs 25 mins). While we're still working through evaluations for other recent models, this is our highest published time horizon to date.