Ornith-1.5: From Self-Scaffolding to Self-Improvement

Kimi Linear: An Expressive, Efficient Attention Architecture

A $500 RL fine-tune of a 9B open model beat frontier models on catalog review

The Little Book of Reinforcement Learning

Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning

LeMario: Training a JEPA World Model on Super Mario Bros

Is One Layer Enough? A Single Transformer Layer Matches Full-Parameter RL Train

Ornith-1.0: self-improving open-source models for agentic coding

Maxproof

DeepSeek-V4 on Day 0: From Fast Inference to Verified RL with SGLang and Miles

Hamilton-Jacobi-Bellman Equation: Reinforcement Learning and Diffusion Models

Tree Search Distillation for Language Models Using PPO

Chess engines do weird stuff

MiniMax M2.5 released: 80.2% in SWE-bench Verified

RLHF from Scratch

Reinforcement Learning from Human Feedback