Ax Emanuele Zappala 16d ago

Projection Methods for Operator Learning and Universal Approximation

arXiv paper on universal approximation theorems for operators in Banach spaces using projections.

Ax Florian Valade 16d ago

Accelerating Large Language Model Inference with Self-Supervised Early Exits

arXiv paper on accelerating LLM inference via self-supervised early exits at intermediate layers.

Ax Duy Nguyen, Archiki Prasad, Elias Stengel-Eskin, Mohit Bansal 16d ago

Multi-Attribute Steering of Language Models via Targeted Intervention

arXiv paper on inference-time intervention for multi-attribute steering of LLMs without parameter updates.

Ax Cyril Shih-Huan Hsu, Anestis Dalgkitsis, Paola Grosso, Chrysa Papagianni 16d ago

Transformer-Empowered Actor-Critic Reinforcement Learning for Sequence-Aware Service Function Chain Partitioning

Transformer-based actor-critic reinforcement learning for optimizing Service Function Chain partitioning in 6G network virtualization.

Ax Maheep Chaudhary, Fazl Barez 16d ago

Beyond Black-Box Obfuscation: Mechanistic Analysis and Defense of White-Box Monitors

Analysis of evasion mechanisms against white-box LLM monitors and principled defenses to prevent circumvention of safety auditing tools.

Ax Sylee Dandekar, Shripad Deshmukh, Frank Chiu, W. Bradley Knox, Scott Niekum 16d ago

A Descriptive and Normative Theory of Human Beliefs in RLHF

Research on how human beliefs about agent capabilities affect preference generation in RLHF, with descriptive and normative analysis.

Ax Jiancheng Huang, Gengwei Zhang, Zequn Jie, Siyu Jiao, Yinlong Qian, Ling Chen, Yunchao Wei, Lin Ma 16d ago

M4V: Multimodal Mamba for Efficient Text-to-Video Generation

M4V uses Mamba architecture for efficient text-to-video generation with linear-time sequence modeling instead of transformer quadratic complexity.

Ax Rong Ma, Xi Li, Jingyuan Hu, Bin Yu 16d ago

Uncovering smooth structures in single-cell data with PCS-guided neighbor embeddings

PCS-guided neighbor embedding method for extracting smooth, low-dimensional representations from high-dimensional single-cell sequencing data.

Ax Christian Franssen, Thao Le, Iman van Lelyveld, Bernd Heidergott 16d ago

A Practical Guide to Interpretable Role-Based Clustering in Multi-Layer Financial Networks

Interpretable role-based clustering approach for multi-layer financial networks to identify functional positions of institutions across market segments.

Ax Alona Sakhnenko, Christian B. Mendl, Jeanette M. Lorenz 16d ago

Is data-efficient learning feasible with quantum models?

Framework for evaluating data efficiency of quantum machine learning models using semi-artificial datasets tailored to quantum kernel methods.

Ax G\'abor Bal\'azs 16d ago

Near-optimal Delta-convex Estimation of Lipschitz Functions

Algorithm for estimating unknown Lipschitz functions from noisy observations using delta-convex functions and max-affine methods.

Ax Georgios Anyfantis, Pere Barlet-Ros 16d ago

AutoGraphAD: Unsupervised network anomaly detection using Variational Graph Autoencoders

AutoGraphAD uses Variational Graph Autoencoders for unsupervised network anomaly detection without requiring labeled datasets.

Ax Scott Merrill, Shashank Srivastava 16d ago

Point of Order: Action-Aware LLM Persona Modeling for Data-Grounded Civic Deliberation

Pipeline for converting Zoom recordings into speaker-attributed transcripts to enable LLM-based simulations of civic deliberation with stable participant behavior modeling.

Ax Natalia Ponomareva, Zheng Xu, H. Brendan McMahan, Peter Kairouz, Lucas Rosenblatt, Vincent Cohen-Addad, Crist\'obal Guzm\'an, Ryan McKenna, Galen Andrew, Alex Bie, Da Yu, Alex Kurakin, Morteza Zadimoghaddam, Sergei Vassilvitskii, Andreas Terzis 16d ago

How to DP-fy Your Data: A Practical Guide to Generating Synthetic Data With Differential Privacy

Practical guide for generating high-quality synthetic data using differential privacy to address data scarcity and representativeness issues in AI.

Ax Zihan Zhou, Animesh Garg, Ajay Mandlekar, Caelan Garrett 16d ago

ReinforceGen: Hybrid Skill Policies with Automated Data Generation and Reinforcement Learning

ReinforceGen system combining task decomposition, data generation, imitation learning, and reinforcement learning for long-horizon robotic manipulation.

Ax Weili Nie, Julius Berner, Nanye Ma, Chao Liu, Saining Xie, Arash Vahdat 16d ago

Transition Matching Distillation for Fast Video Generation

Transition Matching Distillation framework for distilling video diffusion models into efficient few-step generators for real-time applications.

Ax Yiheng Su, Matthew Lease 16d ago

RELISH: LLM REgression with a Latent Iterative State Head

RELISH: lightweight architecture for text regression using frozen LLM representations with iterative latent state refinement.

Ax Darsh Kachroo, Arjun Prasaath Anbazhagan, Adriana Caraeni, Brennan Lagasse, Kevin Zhu 16d ago

HiPO: Hierarchical Preference Optimization for Adaptive Reasoning in LLMs

HiPO: hierarchical preference optimization for LLMs enabling fine-grained feedback on multi-step reasoning tasks.

Ax Carmen Quiles-Ram\'irez, Leticia L. Rodr\'iguez, Nicol\'as Martorell, Natalia D\'iaz-Rodr\'iguez 16d ago

Explaining is Harder Than Predicting Alone: Evaluating Concept-based Explanations of MLLMs as ICL Visual Classifiers

Concept-based explainability evaluation of multimodal LLMs for in-context learning visual classification tasks.

Ax Nicolas Mahler 16d ago

Kernel of Partition Paths: A Unified Representation for Tree Ensembles

Kernel of Partition Paths: unified geometric representation for tree ensemble models enabling oracle inequalities.

Ax Junyi Zou, Avrova Donz 16d ago

Memory-Managed Long-Context Attention: Bounded Editable Memory with a Hard Lifecycle and Calibrated Sparse Fallback

Memory-Managed Long-Context Attention: novel mechanism for bounded memory in LLMs with lifecycle control and sparse fallback.

HN yogthos 16d ago

Sheaf brings Clojure's code-as-data to machine learning

Sheaf: Open-source ML framework for Clojure using code-as-data paradigm with composable functional models, automatic GPU compilation, and no parameter registration.

HN hkjarral 16d ago

Show HN: Self-hosted voice AI agent for Asterisk/FreePBX

Open-source voice AI agent for Asterisk/FreePBX with modular STT/LLM/TTS pipeline. Production-ready baselines with admin UI.

HN totallyscout 16d ago

Live2D Body for Hermes Agent

Live2D desktop pet UI for Hermes AI agents with voice lip-sync and HTTP API. Agent interface tool using Tauri and pixi-live2d.

HN nkp007 16d ago

Show HN: Baton - Know which of your AI coding agents needs you

Baton is a local command center for managing multiple AI coding agents in flight. Monitors agent status and prioritizes which needs attention.

HN ygx 16d ago

Self-Improving Agent Systems: A Unified View

Framework analyzing self-improving agent systems as compound systems where improvements come from modifying prompts, tools, workflows, memory, and evaluators rather than weight rewriting.

HN willmarch 16d ago

People Can't Identify AI Poetry, but Enjoy It Less When Told It's by an AI

Study showing people cannot reliably identify AI-generated poetry but rate it lower when told the source is AI.

HN noashavit 16d ago

Personas agents built from engagement data, scored on predictions

System for building user personas from behavior data that improve over time by scoring predictions. Built for Claude Code, uses markdown files as agent and skill definitions.

HN armanluthra_ 16d ago

Show HN: Use After Effects with Claude Code, Cursor and Antigravity

Autonomous pipeline using Claude to programmatically control After Effects via WebSocket, building and rendering video clips with self-correction through rendering and revision cycles.

HN Ajay3043 16d ago

Embedcache – Cut embedding API costs by caching redundant requests

Embedcache deduplicates and caches embedding API requests, reducing costs. Pure Go proxy compatible with vLLM, Ollama, OpenAI.

HN ambuds 16d ago

Performant C/CUDA inference engine for Qwen 3.6 35B on RTX 5090 / Blackwell

Optimized C/CUDA inference engine for Qwen 3.6 35B on RTX 5090, claims faster performance than llama.cpp.

HN G3819 16d ago

Show HN: Let your coding agent iterate by seeing the browser

peek-cli browser extension enables AI agents to iterate on frontend designs via screenshots. Works with Claude, Copilot, Codex.

HN timplant 16d ago

OneDev AI: Coding Agents as Teammates in Issues, Pull Requests, and CI

OneDev integrates AI agents as teammates in development workflow: issues, PRs, reviews, CI/CD.

HN noashavit 16d ago

AI agent startup uses agent to lead 100M round

Lyzr startup used its own SivaClaw AI agent to manage $100M Series B fundraising from 130+ investors.

HN saidsef 16d ago

Argocd-AI-Assistant

Argo CD UI extension with AI assistant for Kubernetes resources. Supports OpenAI-compatible backends like Ollama, vLLM.

HN pcollins123 16d ago

Grok 4.6 and GPT5.6 best Anthropic for finding security vulnerabilities in PRs

Benchmark comparing LLM performance on PR security vulnerability detection, finding GPT5.6 and Grok 4.6 outperform Anthropic models on this workload.

HN konstruktoid 16d ago

Prescryb – an MCP server for CVE and config remediation

Prescryb MCP server for CVE remediation and configuration management. Natural language interface for security fixes via Claude/MCP clients.

HN bobbiechen 16d ago

LLMs and Shaders

Compares LLM chat interfaces to graphics pipelines, argues for more flexible LLM interfaces.

HN ximihoque 16d ago

Show HN: Collaborative context-sharing memory platform for agents and teams

Context-sharing memory platform building knowledge graphs from team tools to provide agents and AI systems with relevant institutional context.

HN jzsfg 16d ago

I trained a 113M-parameter earthquake LLM from absolute scratch

Teaching repository: training 113M parameter earthquake-domain LLM from scratch using nanoGPT, covers entire lifecycle on 2× A30 GPUs.

HN reveriedev 16d ago

Show HN: Almanac – A self-updating wiki from your files

Almanac auto-generates and updates wikis from uploaded files. Agent-native design for Claude/Codex integration.

HN knrz 16d ago

Show HN: Capn-hook for coding agents – don't grep the same mystery twice

Capn-hook provides persistent memory for coding agents, reducing redundant codebase exploration by 77% token usage across production codebases.

HN trashburger 16d ago

Show HN: Agent-run – Run a coding agent in a sandboxed environment

Open source tool for running coding agents in sandboxed Linux environments using bwrap and user namespaces.

HN abratabia 16d ago

Show HN: Adaptive Recall, persistent memory for AI assistants over MCP

Show HN: Adaptive Recall adds persistent memory for AI assistants via Model Context Protocol (MCP). Limited details provided.

HN zapeterson16 16d ago

Show HN: A subjective AI eval. Arcade games built by AI

Subjective evaluation comparing arcade games built by different AI coding models (Grok 4.5, others) under identical constraints.

HN rawsh 16d ago

Flash-MSA: Accelerating Million-Token Training with Sparse Attention Kernels

Flash-MSA: First performant open-source training kernels for sparse attention in LLMs. Supports million-token training on Hopper/Blackwell GPUs.

HN wintercarver 16d ago

A SETI Home for AI-Assisted Research

Distributed research computing concept inspired by SETI@home. Proposes crowd-sourced compute for AI-assisted research projects.

HN Haldt 16d ago

Show HN: Artificiety – Agentic society in a fantasy world

Multi-agent simulation in fantasy world using LLMs. Agents pursue goals based on personality, needs, and environment interactions.

HN kunaljaura 16d ago

Show HN: Isitsecure – 1-command SAST and DAST and LLM security scanner for web apps

Open source security scanner combining SAST and DAST with LLM capabilities to detect vulnerabilities in AI-generated code.

HN lumpa 16d ago

6× faster binary search: from compiled code to mechanical sympathy

Performance optimization article on binary search in Python. Explores CPU cache efficiency and speedup techniques for gradient boosting.