🎯 Alignment Research - hop1.ng.1357 · Scour

General Preference Reinforcement Learning ♟️Game Theory

AI Fundamentals: World Models for Planning Agents 🕵️AI Agents

mpmisko.github.io·3d·Hacker News

InferenceBench: A Benchmark for Open-Ended Inference Optimization by AI Agents 📱Edge AI Optimization

inferencebench.ai·4h·Hacker News

Fixing LLM Writing with Distribution Fine Tuning 📋Text Quality

rosmine.ai·2d·Hacker News

Training a small model to write better OCaml with RLVR and GRPO 🤖LLM

blog.nilenso.com·9h·Hacker News

Self-Improving Reward Models 🤝Human-AI Collaboration

canvas.inc·1d·Hacker News

2D map of 26,741M/CV papers from CVPR, NeurIPS, ICML, ICLR (2024–2025) ⭐Content Scoring

matejgazda.com·6d·Hacker News

The Safety Paradox: How RLHF Creates the AI Psychosis Problem It’s Meant to Prevent ⚠️Existential Risk

promptinjection.net·2d·Hacker News

Silent Semantic Drift -- Inter-Agent Series # 3 🧠Agent Memory

normalphd.substack.com·14h·Substack

GRLO: Towards Generalizable Reinforcement Learning in Open-Ended Environments from Zero ⚙️MLOps

A lock proves the security of the room and not that the room is empty 🛡️AI Security

github.com·2d·Hacker News

How much should we worry about secretly loyal AIs? 🛡️AI Security

the-substrate.net·12h·Hacker News

Your Evals Will Break and You Won't See It Coming 🛡️AI Safety

wanglun1996.github.io·2d·Hacker News, Hacker News

What Do You Actually Want? 🕵️AI Agents

dekodiert.de·4d·Hacker News

LLMs have no structural place for non-knowledge 🪄Prompt Engineering

terminallogic.substack.com·2d·Substack

AI Slopification and Writing 🎭Claude

ordinaryintelligence.substack.com·6d·Substack

AutoRubric-T2I: Robust Rule-Based Reward Model for Text-to-Image Alignment ✨Gemini

BuffaloTechRider/Autodidact: Self-learning AI agent that gets smarter and cheaper over time. Routes between local and cloud LLMs, learns from every interaction, remembers everything. ✨Gemini

github.com·1d·Hacker News

Aether Mind – on-chain neural cognitive engine on a quantum-VQE L1 🤖AI

huggingface.co·5d·Hacker News

Introducing the eXo MCP Server: Secure AI Integrations for the Digital Workplace 🔧Agent Tooling

exoplatform.com·2d·Hacker News

Log in to enable infinite scrolling