Ax Masum Hasan, Junjie Zhao, Ehsan Hoque 10d ago

HAL: Inducing Human-likeness in LLMs with Alignment

HAL framework aligns LLMs to conversational human-likeness using interpretable, data-driven methods rather than relying solely on scale or broad supervised training.

Ax Valentine Olanubi (University of Alabama, Department of Mathematics), Phineas Agar (University of Alabama, Department of Mathematics), Brendan Ames (University of Southampton, School of Mathematical Sciences) 10d ago

Provably Finding a Hidden Dense Submatrix among Many Planted Dense Submatrices via Convex Programming

Convex programming approach to finding hidden dense submatrices among multiple planted dense submatrices with combinatorial optimization applications.

Ax Fengyuan Liu, Jay Gala, Nilaksh, Dzmitry Bahdanau, Siva Reddy, Hugo Larochelle 10d ago

BRIDGE: Predicting Human Task Completion Time From Model Performance

BRIDGE maps model benchmark performance to human task completion time via psychometric framework, scaling AI capability evaluation without extensive annotations.

Ax Emilio Cruciani, Sebastian Forster, Antonis Skarlatos 10d ago

Incremental (k, z)-Clustering on Graphs

Incremental (k,z)-clustering solves graph clustering problems under adversarial edge updates with explicit solution maintenance in dynamic settings.

Ax Xiaoxi Li, Wenxiang Jiao, Jiarui Jin, Haoxuan Li, Hao Wang, Shijian Wang, Guanting Dong, Jiajie Jin, Yinuo Wang, Yuan Lu, Ji-Rong Wen, Zhicheng Dou, Zhouchen Lin 10d ago

OmniGAIA: Towards Native Omni-Modal AI Agents

OmniGAIA is a benchmark for evaluating omni-modal AI agents with vision, audio, and language capabilities for complex reasoning and tool usage tasks.

Ax Somjit Roy, Pritam Dey, Bani K. Mallick 10d ago

VaSST: Variational Inference for Symbolic Regression using Soft Symbolic Trees

VaSST introduces a probabilistic framework for symbolic regression using variational inference and soft symbolic trees with uncertainty quantification for scientific discovery.

Ax Drew Prinster, Clara Fannjiang, Ji Won Park, Kyunghyun Cho, Anqi Liu, Suchi Saria, Samuel Stanton 10d ago

Conformal Policy Control

Conformal Policy Control uses safe reference policies to regulate untested agent behaviors, balancing exploration and safety constraints in high-stakes environments.

Ax Yinpeng Wu, Yitong Chen, Lixiang Wang, Jinyu Gu, Zhichao Hua, Yubin Xia 10d ago

FlexServe: A Fast and Secure LLM Serving System for Mobile Devices with Flexible Resource Isolation

FlexServe enables privacy-preserving LLM inference on mobile devices using ARM TrustZone for secure model weight and user data protection against OS-level attacks.

Ax Eden Saig, Tamar Garbuz, Ariel D. Procaccia, Inbal Talgam-Cohen, Jamie Tucker-Foltz 10d ago

Adaptive Contracts for Cost-Effective AI Delegation

Adaptive contracts framework for cost-effective AI delegation balancing evaluation noise and costs in pay-for-performance tasks.

Ax Zekun Qian, Wei Feng, Ruize Han, Junhui Hou 10d ago

COVTrack++: Learning Open-Vocabulary Multi-Object Tracking from Continuous Videos via a Synergistic Paradigm

COVTrack++: open-vocabulary multi-object tracking using continuous video data and foundation models for novel objects.

Ax Ye Mao, Weixun Luo, Ranran Huang, Junpeng Jing, Krystian Mikolajczyk 10d ago

RGB-Pointmap Pretraining for Unified 3D Scene Understanding

UniScene3D: transformer framework for 3D scene understanding via RGB-Pointmap alignment with CLIP pretraining.

Ax Carles Domingo-Enrich, Jiequn Han 10d ago

Adjoint Matching through the Lens of the Stochastic Maximum Principle in Optimal Control

Adjoint matching framework for learning optimal controls in diffusion/flow models via stochastic optimal control theory.

Ax Jiacheng Liu, Xiaohan Zhao, Xinyi Shang, Zhiqiang Shen 10d ago

Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems

Analysis of Claude Code agentic system architecture with comparison to OpenClaw and Hermes Agent identifying design principles.

Ax Zenghui Zhou, Man Li, Xiaoke Fang, Xinyi Zhou, Weibin Lin, Zheng Zheng 10d ago

LGMT: Logic-Grounded Metamorphic Testing for Evaluating the Reasoning Reliability of LLMs

LGMT: logic-grounded metamorphic testing framework for evaluating LLM reasoning robustness using first-order logic.

Ax Minhao Yao, Ruoyu Wang, Xihong Lin, Lin Liu, Zhonghua Liu 10d ago

Gradient-Flow Optimization as Dynamic Random-Effects Inference: Testing and Early Stopping with Applications to Deep Learning

Statistical framework viewing gradient-flow optimization as random-effects inference with applications to early stopping in deep learning.

Ax Daisuke Inoue, Dante Kalise, Mathieu Lauri\`ere 10d ago

Nonlocal Mean Field Schr\"{o}dinger Bridge with Learned Interactions

Mean-field Schrödinger bridge with learned nonlocal interactions for efficient large-scale population dynamics.

Ax Jessica Rodrigues, Angelo Salatino, Gard Jenset, Scott Hale 10d ago

Representing Research Attention as Contextually Structured Flows

Framework for representing research attention as contextually structured flows rather than isolated volume metrics.

Ax Hyunjin Cho, Youngji Roh, Jaehyung Kim 10d ago

Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanisms

Unsupervised feature discovery aligning semantics and mechanisms for auditing internal LLM computations via mechanistic interpretability.

Ax Lisa Herzog, Oliver D\"urr, Pascal B\"uhler, Hakim Baazaoui, Julian Dese\"o, Susanne Wegener, Beate Sick 10d ago

Estimating Individualized Treatment Effects in Acute Ischemic Stroke with Causal Transformation Models (TRAM-DAG): A Multi-Centre Observational Study with External RCT Validation

Multi-center observational study using causal models to estimate individualized treatment effects in acute ischemic stroke.

Ax Xinwei Qiang, Yifan Hu, Shixuan Sun, Jing Yang, Han Zhao, Chen Chen, Yu Feng, Jingwen Leng, Minyi Guo 10d ago

GF-DiT: Scheduling Parallelism for Diffusion Transformer Serving

GF-DiT: dynamic parallelism scheduler for efficient diffusion transformer serving with heterogeneous workloads.

Ax Swapnil Manna, Timothy J. Rogers, Lawrence Bull 10d ago

Orthogonal Discrepancy Kernels for Learning with Partial Physics

Semi-parametric framework combining physics-based models with Gaussian process regression for nonlinear system identification.

Ax Yinpeng Wu, Yitong Chen, Lixiang Wang, Jinyu Gu, Zhichao Hua, Yubin Xia 10d ago

FlexServe: A Fast and Secure LLM Serving System for Mobile Devices with Flexible Resource Isolation

FlexServe: system for secure LLM inference on mobile devices using ARM TrustZone hardware isolation.

Ax Alia Tarek, Hamsa Saberr, Hamza Elghonemy, Youssef Afify, Tamer Basha, Omair Shahzad Bhatti, Abdulrahman M. Selim, Hasan Md Tusfiqur Alam, Daniel Sonntag 10d ago

TRACE: A Concept Bottleneck Model for Longitudinal 3D Glioblastoma Response Assessment

TRACE: concept bottleneck model for interpretable glioblastoma response classification from longitudinal MRI data.

Ax Shuwen Chai, Qiaosen Wang 10d ago

Sample Complexities of Estimating Gumbel--Max Watermark Proportions with and without Reduction to Pivotal Statistics

Statistical framework for estimating proportion of LLM-generated text in mixed documents using Gumbel-max watermarking.

HN wslh 10d ago

GPT-5.5-Cyber built a zlib fuzzing lab in a day

OpenAI GPT-5.5-Cyber used for security fuzzing in open-source projects. Collaboration to find/patch bugs before malicious use.

HN emersonmacro 10d ago

Don't Train the Model, Evolve the Harness

Stub article title with no content. Insufficient information.

HN keepamovin 10d ago

Ask HN: Once you make your money from vibe coding innumerable products, then?

Personal blog about product creation strategy and marketing. Off-topic discussion.

HN akarshhegde18 10d ago

Ask HN: How do you get your open-source product good traction?

Discussion thread asking for open-source product growth strategies. Minimal content.

HN snoka 10d ago

FlickerScope – are your LEDs giving you a headache? stop guessing

Browser-based tool for analyzing LED flicker frequency from video. Client-side processing, no uploads.

HN Bender 10d ago

Why AI Is Collapsing: How China Is Winning. [video]

Video title only. Geopolitical AI analysis claim without content.

HN 1vuio0pswjnm7 10d ago

AI is 'not smart' so what's next in artificial intelligence?

Yann LeCun interview on AI limitations beyond current systems. Discussion of physical world understanding gap.

HN alonsovm44 10d ago

I vibe coded a programming language, but I'd rather learn C the old way

Personal account of building programming language with AI vs learning C traditionally. Narrative about AI limitations.

HN sporkl_l 10d ago

Some Basic LLM Etiquette

Opinion piece on LinkedIn professionalism and AI-generated content tone. Social commentary.

HN Bender 10d ago

How AI Became More Expensive Than the Workers It Replaced [video]

Video title only. AI cost comparison claim without details.

HN linzhangrun 10d ago

Ask HN: Best Local LLM Setup for a 128GB M4 Max Mac Studio?

Ask HN thread requesting local LLM setup advice for high-end Mac. No responses.

HN sherlockxu 10d ago

Open Source LLM Statistics and Trends (2026)

Statistics report on open-source LLM ecosystem growth in 2026. Covers usage, performance, inference costs, and market trends.

HN lioeters 10d ago

Epistemic Heat Death and the Signal-to-Noise Ratio of the Global Web

Theoretical framework analyzing information collapse as AI-generated content dominates the web. Proposes 'epistemic heat death' concept.

HN mxfeinberg 10d ago

TurboQuant can reduce vector index size by 10x at 100M Row Scale

TurboQuant vector index compression reduces size by 10x at scale. Open-source implementations for KV-cache compression in LLMs.

HN PaybackTony 10d ago

Show HN: Drop Flap Boards for All

Live Hacker News drop flap board visualization tool. Self-hosted app for creating custom news boards.

HN KingKunta 10d ago

LawZero: Safety from Honesty in a Disinterested AI Predictor

arXivLabs framework announcement with no technical content about the actual research.

HN sudo_cowsay 10d ago

What is agentic AI today, and what do we want it to be?

Analysis of agentic AI deployment trends. MIT research on current state and future potential of automated software agents.

HN verdverm 10d ago

Sebastian Mallaby: The AI Race Nobody Can Win, Foreign Affairs Interview [video]

Video title only. Foreign policy interview summary unavailable.

HN bmcdresson 10d ago

GLM-5.2: The Open-Source Chinese Model Challenging Claude at One-Fifth the Cost

GLM-5.2: open-source Chinese frontier LLM matching Claude/GPT performance at 1/5 cost under MIT license.

HN vaishcodescape 10d ago

Using AI Agents with Databases

Data-Spear: autonomous SQL agent for PostgreSQL that plans queries, verifies results, and cites sources.

HN matt_d 10d ago

Stealing 50 Years of Database Ideas for AI Agents

OneWill system for safely sandboxing autonomous agents using database-inspired write-ahead logging and state isolation.

HN Anon84 10d ago

Learning to Replicate Expert Judgment in Financial Tasks

Research on using LLMs to replicate expert judgment in financial decision-making tasks.

HN lambda_foo 10d ago

OCaml 5.5 Released

OCaml 5.5.0 release notes covering functional programming language updates.

HN matt_d 10d ago

Empirical Computation: Prompting versus Programming [pdf]

PDF title about comparing prompting versus programming approaches. No content provided.

HN healsdata 10d ago

The $1.3M theft that exposed AI's blind spot

Cybersecurity article about AI blind spots in threat detection.

HN ctrlnode-ai 10d ago

We Ran a Complex Task – A LangChain Repo Analysis with Claude Fable Models

Comparative analysis of Claude Fable models (Opus, Sonnet, Haiku) on complex engineering task using multi-agent approach.