Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance Paper • 2511.13254 • Published Nov 17, 2025 • 136
Training Long-Context, Multi-Turn Software Engineering Agents with Reinforcement Learning Paper • 2508.03501 • Published Aug 5, 2025 • 59
view article Article SmolVLA: Efficient Vision-Language-Action Model trained on Lerobot Community Data +7 Jun 3, 2025 • 305
Quartet: Native FP4 Training Can Be Optimal for Large Language Models Paper • 2505.14669 • Published May 20, 2025 • 78