AI/ML arXiv cs.AI

Self-Improvements in Modern Agentic Systems: A Survey

A comprehensive survey on self-improving autonomous agents, providing a system-level framework for how agents convert experience into capability gains.

AI/ML arXiv cs.AI

Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools

Introduces a Context-Augmented Prompting framework that uses GNN experts to improve molecular property prediction in small language models.

AI/ML arXiv cs.AI

Oracle Agent Memory as an Enterprise Memory Substrate for Long-Horizon AI Agents

Explores Oracle Agent Memory as a database-native substrate for long-horizon AI agents to manage state, user preferences, and procedural knowledge.

AI/ML arXiv cs.AI

Learning Safe Agent Behaviour from Human Preferences and Justifications via World Models

Presents DROPJ, a human-centered method for training safe agent policies using world models and justified human preferences.

Software Engineering arXiv cs.AI

CayleyR: Solving the TopSpin puzzle via cycle intersection

Introduces cayleyR, an R package for solving permutation puzzles like TopSpin by detecting cycle intersections in Cayley graphs, featuring C++ and Vulkan GPU acceleration.

AI/ML arXiv cs.AI

Networked Intelligence: Active Shared Context Graphs for Human-AI Team Science

Introduces Mycelium, an active shared workspace for human-AI team science that routes shared context and hypotheses across a network of researchers and agents.

Software Engineering Hacker News

If you want to create a button from scratch, you must first create the universe

A discussion or essay on the inherent complexity and dependencies involved in building even simple UI elements from first principles.

AI/ML Hacker News

An Interactive Map of AI

A visual exploration and mapping of the diverse and rapidly evolving landscape of artificial intelligence technologies.

Software Engineering Hacker News

Making 768 servers look like 1

A technical exploration of server clustering and abstraction techniques to make a massive fleet of servers appear as a single unit.

Open Source Hacker News

Open Source, Free Tier Capable Whispr Using Cloudflare AI

Implementation of a free-tier capable transcription service using Cloudflare AI, focusing on open-source accessibility.

Other Hacker News

My Throw Decides My Aim

A project exploring the intersection of physical motion (throwing) and digital aiming mechanisms, likely in a gaming or simulation context.

AI/ML Hacker News

Accelerating Block Low-Rank Foundation Model Inference on MemoryConstrained GPUs

Research on accelerating foundation model inference on memory-constrained GPUs using block low-rank approximations.

Other Hacker News

1,300 Beautiful Wildlife Illustrations from the 19th Century Now Restored

The restoration and digital release of 1,300 wildlife illustrations from the 19th century.

Tech Business/VC TechCrunch

Lululemon backs nylon recycling startup Syntetica in $30M Series A

Nylon recycling startup Syntetica raises $30M in Series A funding with backing from Lululemon.

Tech Business/VC TechCrunch

Applied Computing wants to give oil and gas operators an AI model for the entire plant

Applied Computing raises $20M Series A to develop foundation AI models specifically for the oil and gas industry.

AI/ML arXiv cs.AI

OriginBlame: Record- and Token-Level Data Provenance for AI Training Datasets

Introduction of OriginBlame, a system for record- and token-level data provenance in AI training to enable precise data removal and unlearning.

Other Hacker News

Guerrilla London Bus Ads Mock Kylie Jenner's Meta Glasses Campaign

A guerrilla marketing campaign in London mocks Meta's glasses campaign featuring Kylie Jenner.

Hardware/Chips Hacker News

High-Bandwidth Flash offers efficient storage for model weights

Discussion on using high-bandwidth flash storage to efficiently store and load model weights for AI models.

Hardware/Chips Hacker News

Fleet: Hierarchical Task-Based Abstraction for Megakernels on Multi-Die GPUs

Introduction of Fleet, a hierarchical task-based abstraction for managing megakernels on multi-die GPUs.

Software Engineering Lobste.rs

Linus Torvalds on LLM usage in kernel development

Linus Torvalds discusses the use and limitations of LLMs in the development of the Linux kernel.