Category: Model Optimization

• 35B-Parameter Science Model Rivals Trillion-Parameter Giants: 'Intern' Science Model Intern-S2-Preview Open-Sourced
• ICML 2026 | Rejecting Brute Force, PRISM Framework Enables Efficient Test-Time Scaling for dLLMs
• Qwen-Scope: Seeing Through the 'Hidden Thoughts' of Large Models
• Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought
• Scaling Laws for Looped Transformers
• Long Context Reduced by 60% + 95% Sparsity: A Double Breakthrough Today Sets New Records in Inference Efficiency
• No Reinforcement Learning Needed! Apple's 'Simple Self-Distillation' Achieves Self-Evolution for Coding Models
• Multimodal Video Streaming Inference Efficiency Boosted by 56%: Unveiling TWW's Segment-Level Dynamic Memory Mechanism
• Are LLM RL Training Trajectories Actually Linear? Miaow Lab's Latest Work: Directly 'Predict' Future Models Without Further Training!
• A New Revolution in Reward Models! SWIFT Reads "Inner Voice" Instead of Text, Creating a Faster, Stronger, and More Cost-Effective AI Judge
• Is Your Model's Attention Drifting? RUC and Tsinghua University Introduce LeaF: Pruning Distracting Tokens for Focused Learning
• Kimi K2's Key Training Technique: QK-Clip!
• Mianbi MiniCPM4: 3x Inference Speed, Outperforming Same-Size Qwen3, Putting Pressure on Alibaba
• SLOT: Sample-Specific Inference Optimization Tool Arrives, Boosting Accuracy by 10% Without SFT or RL
• Ushering in the Era of On-Device Long Text! OpenBMB's New Architecture Boosts MiniCPM up to 220x Faster
• Deep Learning: Mamba Core Author's New Work Replaces DeepSeek's Attention Mechanism, Designed for Inference