• Diffusion Models Are Real-Time Game Engines
• Distributed Training Over-the-Internet
• Reward Modeling as Next-Token Prediction
• A Law of Next-Token Prediction in LLMs
• The Mamba in the Llama
• Physics of Language Models
• BaichuanSEED
• Long Context as a New Modality
• PolyRouter
• Training-Free Activation Sparsity in LLMs
• Pluto and Charon
• RAGLAB
• Fire-Flyer AI-HPC
• Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts
• Text2SQL is Not Enough
• Unsupervised-to-Online Reinforcement Learning
• LlamaDuo
• Performance Law of Large Language Models

MLLM/VLM/CV