HN danielmateo773 8d ago

Wan 2.7 – AI Video Generator for Text-to-Video and Image-to-Video

Product announcement for Wan 2.7 AI video generator supporting text-to-video and image-to-video with 4K output.

HN chillax 8d ago

Vulnpocalypse: AI, Open Source, and the Race to Remediate

Analysis of AI-generated spam vulnerability reports overwhelming open source maintainers like cURL's Daniel Stenberg, exploring AI's impact on open source security.

HN prasanthsd 8d ago

Show HN: Shell-MCP A persistent terminal for AI- CD, env vars,and nvm carry over

Open source tool providing persistent terminal context for AI agents. Maintains directory, environment variables, and nvm state across sessions via MCP protocol.

HN latentframe 8d ago

Scaling AI is now constrained by energy, cooling and physics

Brief headline about energy and cooling constraints on AI scaling without substantive content provided.

HN wasimsk 8d ago

Ask HN: Does Vibe Coding and Prompt Engineering make me an Software Engineer?

Discussion post asking if prompt engineering and LLM-driven development constitutes software engineering. Community debate on AI-assisted coding.

HN vinhnx 8d ago

Let's Talk about LLMs

Headline only, no content provided.

Ax Juan Du, Yueteng Wu, Pan Zhao, Yuze Liu, Min Zhang, Xiaobin Xu, Xinglong Zhang 8d ago

TurboAgent: An LLM-Driven Autonomous Multi-Agent Framework for Turbomachinery Aerodynamic Design

TurboAgent is an LLM-driven multi-agent framework enabling autonomous end-to-end turbomachinery aerodynamic design through coordinated geometry, prediction, optimization stages.

Ax Sungwoo Jung, Seonil Son 8d ago

How Much LLM Does a Self-Revising Agent Actually Need?

Empirical study decomposing LLM-based agent competence to determine which capabilities derive from the language model versus explicit architectural structure.

Ax Jaehyeok Lee, Xiaoyuan Yi, Jing Yao, Hyunjin Hwang, Roy Ka-Wei Lee, Xing Xie, JinYeong Bak 8d ago

Distributional Open-Ended Evaluation of LLM Cultural Value Alignment Based on Value Codebook

DOVE benchmark evaluates LLM cultural value alignment using open-ended generation to address limitations of multiple-choice evaluation formats.

Ax Joy Dutta, Hossien B. Eldeeb, Tu Dac Ho 8d ago

Blockchain and AI: Securing Intelligent Networks for the Future

Survey synthesizing blockchain and AI integration for securing intelligent networks, covering ledger design, detection, and agentic workflows.

Ax Manish Bhatt, Sarthak Munshi, Vineeth Sai Narajala, Idan Habler, Ammar Al-Kahfah, Ken Huang, Joel Webb, Blake Gatto 8d ago

The Defense Trilemma: Why Prompt Injection Defense Wrappers Fail?

Formal proof that continuous wrapper defenses cannot protect LLMs from all prompt injection attacks, characterizing where every defense must fail.

Ax Hanyang Wang, Mingxuan Zhu 8d ago

The Detection-Extraction Gap: Models Know the Answer Before They Can Say It

Empirical study showing 52-88% of chain-of-thought tokens in reasoning models are generated after the answer is already recoverable, revealing the detection-extraction gap phenomenon.

Ax Jiang Zhou, Yunhao Wang, Xing Wu, Tinghao Yu, Feng Zhang 8d ago

WRAP++: Web discoveRy Amplified Pretraining

WRAP++ enhances LLM pretraining through synthetic data rephrasing that captures cross-document relationships beyond single-document web page rewriting.

Ax Eduardo C. Garrido-Merch\'an, \'Alvaro L\'opez L\'opez 8d ago

An Automated Survey of Generative Artificial Intelligence: Large Language Models, Architectures, Protocols, and Applications

Comprehensive survey of generative AI and LLMs covering model families, deployment protocols, and real-world applications as of early 2026.

Ax Zihao Zhao, Yuchen Yang, Anjalie Field, Yinzhi Cao 8d ago

AdaProb: Efficient Machine Unlearning via Adaptive Probability

AdaProb proposes efficient machine unlearning via adaptive probability to address residual information and computational overhead in removing specific data from trained models.

Ax Renhua Ding, Xiao Yang, Zhengwei Fang, Jun Luo, Kun He, Jun Zhu 8d ago

Invisible to Humans, Triggered by Agents: Stealthy Jailbreak Attacks on Mobile Vision-Language Agents

Study of stealthy visual jailbreak attacks on mobile Vision-Language agents that exploit discrepancies between LVLM perception and human vision without user detection.

Ax Carla Crivoi, Radu Tudor Ionescu 8d ago

Machine Unlearning in the Era of Quantum Machine Learning: An Empirical Study

First empirical study of machine unlearning in variational quantum circuits and quantum-augmented neural networks, adapting classical unlearning methods to quantum settings.

Ax Xiang Li, Xueheng Li, Yu Wang, Xuanhua He, Zhangchi Hu, Weiwei Yu, Chengjun Xie 8d ago

Q-Probe: Scaling Image Quality Assessment to High Resolution via Context-Aware Agentic Probing

Q-Probe: Agentic multi-scale probing approach for high-resolution image quality assessment using MLLMs with reinforcement learning for human preference alignment.

Ax Jaden Zhang, Gardenia Liu, Oliver Johansson, Hileamlak Yitayew, Kamryn Ohly, Grace Li 8d ago

Prediction Arena: Benchmarking AI Models on Real-World Prediction Markets

Prediction Arena: Benchmark evaluating AI model decision-making by enabling autonomous trading on live prediction markets with real capital and objective ground truth.

Ax Rui Dong, Zitong Wang, Jiaxing Li, Weihuang Zheng, Youyong Kong 8d ago

BLEG: LLM Functions as Powerful fMRI Graph-Enhancer for Brain Network Analysis

BLEG: Framework combining LLMs with Graph Neural Networks to enhance fMRI brain network analysis by addressing feature sparsity and domain knowledge limitations.

Ax Faezeh Pasandideh, Achim Rettberg 8d ago

LLM-Generated Fault Scenarios for Evaluating Perception-Driven Lane Following in Autonomous Edge Systems

Decoupled offline-online fault injection framework using LLMs to generate diverse fault scenarios for testing autonomous vision systems on edge devices.

Ax Hongjian Zou, Yidan Wang, Qi Ding, Yixuan Liao, Xiaoxin Chen 8d ago

Benchmark Shadows: Data Alignment, Parameter Footprints, and Generalization in Large Language Models

Analysis showing LLMs achieve benchmark gains without broader capability improvements due to benchmark-aligned training data limiting generalization.

Ax Yilong Dai, Shengyu Chen, Xiaowei Jia, Runlong Yu 8d ago

Flow Learners for PDEs: Toward a Physics-to-Physics Paradigm for Scientific Computing

Flow Learners framework for learning PDE solutions combining physics-informed constraints with generative AI paradigm for scalable scientific computing.

Ax Ameen Patel, Felix Lee, Kyle Liang, Joseph Thomas 8d ago

The Role of Emotional Stimuli and Intensity in Shaping Large Language Model Behavior

Empirical study on emotional prompt engineering for LLMs, exploring effects of four distinct emotions at varying intensity levels on model performance and behavior.

Ax Yongzhong Xu 8d ago

The Lifecycle of the Spectral Edge: From Gradient Learning to Weight-Decay Compression

Study decomposing the spectral edge lifecycle during grokking, showing gradient-to-weight-decay transition where edge becomes compression axis critical for model generalization.

Ax Benjamin J. Choi, Melanie Weber 8d ago

Latent Structure of Affective Representations in Large Language Models

Analysis of latent geometric structure in LLM representations through emotion processing, investigating how emotional information is encoded and organized.

Ax Yuyao Wang, Min Yang, Meng Chen, Weiming Huang, Yongshun Gong 8d ago

SCOT: Multi-Source Cross-City Transfer with Optimal-Transport Soft-Correspondence Objective

Cross-city transfer learning using optimal transport for region correspondence and improved prediction in label-scarce cities with incompatible geographic partitions.

Ax Shresth Verma, Arpan Dasgupta, Neha Madhiwalla, Aparna Taneja, Milind Tambe 8d ago

Decisions and Deployment: The Five-Year SAHELI Project (2020-2025) on Restless Multi-Armed Bandits for Improving Maternal and Child Health

Five-year SAHELI project applying restless multi-armed bandits algorithms to optimize healthcare worker resource scheduling for maternal and child health program engagement.

Ax David Golchinfar, Daryoush Vaziri, Alexander Marquardt 8d ago

Playing DOOM with 1.3M Parameters: Specialized Small Models vs Large Language Models for Real-Time Game Control

SauerkrautLM-Doom-MultiVec: 1.3M parameter specialized model for real-time DOOM gameplay outperforming LLMs 92,000x larger, using ModernBERT with hash embeddings and attention pooling.

Ax Niloy Das, Apurba Adhikary, Sheikh Salman Hassan, Yu Qiao, Zhu Han, Tharmalingam Ratnarajah, Choong Seon Hong 8d ago

A Novel Edge-Assisted Quantum-Classical Hybrid Framework for Crime Pattern Learning and Classification

Quantum-classical hybrid framework comparing computational paradigms for crime pattern analysis and classification on imbalanced datasets.

Ax Yucheng Sheng, Jiacheng Wang, Le Liang, Hao Ye, Shi Jin 8d ago

A Graph Foundation Model for Wireless Resource Allocation

Graph foundation model for wireless network resource allocation using deep learning to solve optimization problems more efficiently than classical iterative algorithms.

Ax Fan Zhaowen 8d ago

Event-Centric World Modeling with Memory-Augmented Retrieval for Embodied Decision-Making

Event-centric world modeling framework with memory-augmented retrieval for autonomous agent decision-making that balances computational efficiency with physical groundedness.

Ax Yeran Zhang, Pengwei Yang, Guoqing Wang, Tianyu Li 8d ago

DSPR: Dual-Stream Physics-Residual Networks for Trustworthy Industrial Time Series Forecasting

Physics-residual neural network framework for industrial time series forecasting that combines data-driven learning with physical constraints for non-stationary systems.

Ax Quantong Qiu, Zhiyi Hong, Yi Yang, Haitian Wang, Kebin Liu, Qingqing Dang, Juntao Li, Min Zhang 8d ago

Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference

Context-aware hybrid attention mechanism for LLMs that dynamically allocates between full and sparse attention based on task demands to reduce computational complexity.

Ax Jinhong Lin, Pan Wang, Zitong Zhan, Lin Zhang, Pedro Morgado 8d ago

Data Warmup: Complexity-Aware Curricula for Efficient Diffusion Training

Curriculum learning strategy for diffusion model training that schedules images by complexity to improve training efficiency without modifying model architecture.

Ax Wonseon Lim, Jaesung Lee, Dae-Won Kim 8d ago

Critical Patch-Aware Sparse Prompting with Decoupled Training for Continual Learning on the Edge

Sparse prompting method for continual learning on edge devices with reduced memory and computational overhead.

Ax Yucheng Zhou, Jianbing Shen 8d ago

Accelerating Training of Autoregressive Video Generation Models via Local Optimization with Representation Continuity

Techniques for accelerating autoregressive video generation training while managing error accumulation.

Ax Daniel Nobrega Medeiros 8d ago

Conservation Law Breaking at the Edge of Stability: A Spectral Theory of Non-Convex Neural Network Optimization

Spectral theory analysis of gradient descent optimization in ReLU networks, explaining why non-convex optimization works.

Ax Chenchen Xu, Min Zhou, Tiezheng Ge, Weiwei Xu 8d ago

GAN-based Domain Adaptation for Image-aware Layout Generation in Advertising Poster Design

GAN-based model with domain adaptation for automated layout generation in poster design using new dataset.

Ax Kristina Levina, Nikolaos Pappas, Athanasios Karapantelakis, Aneta Vulgarakis Feljan, Jendrik Seipp 8d ago

Reinforcement Learning with Reward Machines for Sleep Control in Mobile Networks

Reinforcement learning with reward machines for optimizing sleep scheduling in mobile networks.

Ax Yuhao Zhang, Ti John, Matthias Stosiek, Patrick Rinke 8d ago

Bayesian Optimization for Mixed-Variable Problems in the Natural Sciences

Bayesian optimization methods for efficient black-box optimization in mixed-variable scientific problems.

Ax Yucheng Zhou, Dubing Chen, Huan Zheng, Jianbing Shen 8d ago

Multimodal Large Language Models for Multi-Subject In-Context Image Generation

MUSIC multimodal LLM for multi-subject in-context image generation, addressing subject identity preservation in text-to-image synthesis.

Ax Prakul Sunil Hiremath 8d ago

GIRL: Generative Imagination Reinforcement Learning via Information-Theoretic Hallucination Control

GIRL latent world-model framework combining generative models with information-theoretic control for long-horizon model-based RL.

Ax Prakul Sunil Hiremath 8d ago

Regret-Aware Policy Optimization: Environment-Level Memory for Replay Suppression under Delayed Harm

Replay Suppression Diagnostic protocol for safe RL under delayed harm, addressing environment-level memory for safety.

Ax Jiaming Cheng, Duong Tung Nguyen 8d ago

Fast Heterogeneous Serving: Scalable Mixed-Scale LLM Allocation for SLO-Constrained Inference

Constraint-aware heuristics for heterogeneous LLM allocation and serving under latency, accuracy, and budget constraints.

Ax Oleg Platonov, Liudmila Prokhorenkova 8d ago

Cluster Attention for Graph Machine Learning

Cluster Attention mechanism for graph transformers improving receptive field while preserving graph-structure inductive biases.

Ax Grace Jiarui Fan, Chengpiao Huang, Tianyi Peng, Kaizheng Wang, Yuhang Wu 8d ago

SYN-DIGITS: A Synthetic Control Framework for Calibrated Digital Twin Simulation

SYN-DIGITS synthetic control framework for calibrating LLM-based digital twin simulations to match real human behavior.

Ax Peter Amorese, Morteza Lahijanian 8d ago

Learning Markov Processes as Sum-of-Square Forms for Analytical Belief Propagation

Sparse Sum-of-Squares functional modeling framework for analytical belief propagation in Markov process models.

Ax Jeffrey D. Varner, Maria Cristina Bravo, Carole McBride, Thomas Orfeo, Ira Bernstein 8d ago

Validated Synthetic Patient Generation for Small Longitudinal Cohorts: Coagulation Dynamics Across Pregnancy

Stochastic Attention generative framework for synthetic patient generation from small longitudinal clinical cohorts.

Ax Henry C. Conklin, Tom Hosking, Tan Yi-Chern, Julian Gold, Jonathan D. Cohen, Thomas L. Griffiths, Max Bartolo, Seraphina Goldfarb-Tarrant 8d ago

Learning is Forgetting: LLM Training As Lossy Compression

LLM training as lossy compression framework explaining how LLMs learn by retaining task-relevant information from training data.