Latest Articles

• Where Are the Inference Limits of Small Models? Weibo Open-Sources 3B Model Rivaling Top-Tier Closed-Source Giants
  AI & Machine LearningSmall Language ModelsOpen Source AITraining MethodologyModel Benchmarks

  ...
• ACL 2026 | Why Does SFT Always Fail to Learn? Not Every SFT Failure Needs More Epochs! Five "Surgical Scalpels" to Fix SFT
  Machine LearningSupervised Fine-TuningContinual Pre-TrainingACL 2026Training OptimizationLarge Language Models

  ...
• Claude Code Lead Says: I No Longer Prompt AI Directly—What Is Loop Engineering?
  AI-Assisted Software DevelopmentLoop EngineeringAI Agent WorkflowsClaude CodePrompt Engineering

  ...
• The End of Code Review: Coding Agents Supersede Human Inspection
  Software EngineeringArtificial IntelligenceAutomationLarge Language ModelsCode Review

  ...
• Anthropic's First Mythic-Class Claude 5 Officially Unleashed
  Artificial IntelligenceLarge Language ModelsScientific ResearchSoftware EngineeringAI Benchmarks

  ...
• SkillOpt: Microsoft Proposes Self-Evolving Agent Skills
  AI Agent OptimizationMicrosoft ResearchMachine Learning SystemsPrompt EngineeringLLM Skill Evolution

  ...
• Anthropic Participates in New Paper: Why Do Larger Models Learn More? The Answer Lies in Scaling
  Machine Learning ResearchLarge Language ModelsGradient InterferenceTraining DynamicsScaling Laws

  ...
• How to Build a Reliable Agent Memory Framework? UC Berkeley's MemFail Stress-Tests 4 Top Memory Systems, Proving Vector Databases Aren't the Only Answer
  AI Memory SystemsLarge Language ModelsMachine LearningBenchmarkingAgent Architecture

  ...
• Open Source Update | Are LLMs Still Stuck with 'Goldfish Memory'? New Benchmark RHELM Tests the Ceiling of 'Real Long-Term Memory'
  AI BenchmarkLarge Language ModelsMicrosoft ResearchRHELMMemory Benchmark

  ...
• Latest Discovery: AI Large Models Know When They're Being Evaluated
  Artificial IntelligenceAI SafetyAI AlignmentLarge Language ModelsModel Evaluation

  ...
• AI Has No ROI
  AI Bubble & ROI CrisisAI EconomicsInvestment ReturnsTech Industry BubblesGenerative AI Costs

  ...
• DeepMind Unveils Co-Scientist Multi-Agent System for Research Hypothesis Generation
  Artificial IntelligenceMulti-Agent SystemsGemini AIGoogle DeepMindScientific Research

  ...
• ChatGPT Memory System Upgrade Launches, Cross-Conversation Context Now Available
  Artificial IntelligenceChatGPTAI AssistantsCross-Conversation ContextMemory Systems

  ...
• Same Day: Hinton Says AI Has Consciousness, Anthropic Says Recursive Self-Improvement Has Arrived
  Artificial IntelligenceAI ConsciousnessLabor DisplacementAI SafetyRecursive Self-Improvement

  ...
• Tencent Open-Sourced Agent Memory, So AI Can Truly Remember You
  Artificial IntelligenceOpen Source SoftwareCloud ComputingDatabase TechnologyMemory Systems

  ...
• Spring Founder Rod Johnson Returns to Frontline AI Frameworks: 'This Is the Last Generation of Frameworks Hand-Chosen by Humans'
  Artificial IntelligenceSoftware DevelopmentProgramming LanguagesOpen SourceEnterprise Technology

  ...
• Google and Cornell's New Research: The Next Step for LLMs Is Learning How to 'Sleep' Well
  AI & Machine LearningLarge Language ModelsTechnologyResearchDeep Learning

  ...
• Qwen3.7-Plus: A Multimodal Agent That Sees, Codes, and Operates Screens—Edge Closer to Practical Use
  Artificial IntelligenceLarge Language ModelTechnology NewsAI AgentMultimodal AI

  ...
• Reflections on the White House's Latest AI Regulation Executive Order
  AI RegulationTrump AdministrationUS-China Tech CompetitionCybersecurityFrontier Models

  ...
• Meta Skill Has Just Arrived
  Artificial IntelligenceAI AgentsTechnology ProductsLarge Language ModelsOpen Source Software

  ...