LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches Paper • 2610.06647 • Published Oct 5 • 35
STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization Paper • 2609.38169 • Published Sep 29 • 37
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness Paper • 2609.08183 • Published Sep 8 • 70