Category: Machine Learning
- An Excellent New Systematic Survey of Self-Evolving Agents
- The Father of GPT Throws AI Back to 1930: Never Saw a Line of Code, Yet 'Invented' Python!
- Skills-Driven Reasoning Paradigm: Tsinghua & Peking University Propose TRS, Saving 59% Tokens Without Accuracy Drop
- Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought
- Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond
- Scaling Laws for Looped Transformers
- NUS, Fudan, and Tsinghua: The First Systematic Survey on Large Model Latent Spaces
- Southeast University's Geng Xin Team: Models Don't Fail Due to Inability, But 'Crowded-Out Capacity' | CVPR 2026
- The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
- Autogenesis: A Self-Evolving Agent Protocol
- The World's Most Notorious Forum Uncovered AI's Most Crucial 'Thinking' Ability
- FrontierSWE
- Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All
- Cognition | Introducing SWE-Check: 10x Faster Bug Detection
- Li Fei-Fei's Team Is Tackling This: From Entropy to Mutual Information, RAGEN-2 Reshapes Reasoning Quality Standards, Preventing AI Agents from Becoming 'More Trained, More Templated'
- Meta Bets on Neural Computers: Is the Next-Gen Computer the Model Itself?
- Frozen Weights Are the Enemy of AI Progress! DeepMind's Top Researcher: The Key to AI Self-Improvement Lies in Evaluation, Drawing from Formal Verification! Expert Models Are Stepping Stones to Generalized AGI!
- The advisor strategy: Give agents an intelligence boost
- AI Doesn't Need to Remember Everything; It Needs to Learn How to Learn: This Memory Revolution Enables Deep Research Agents to Think
- No Reinforcement Learning Needed! Apple's 'Simple Self-Distillation' Achieves Self-Evolution for Coding Models
- Composer 2 Technical Report
- GLM-5.1: Towards Long-Horizon Tasks
- Meta-Harness Supercharges Haiku's Performance, Even Rivalling Opus!
- Chandra OCR 2 Goes Open Source! Scores 85.9 on Official Benchmarks, Crushing GPT-4o's 69.9
- Meta-Harness: Stanford's Latest Harness Paper Earns Praise from Lin Junyong