Latest Articles

Can Large Models Handle Precision Work Too?! MIT Top Conference Paper Teaches AI to Operate Industrial CAD Software
VideoCADMITAI AgentsCAD SoftwareNeurIPS 2025
...
Achieve Cluster Efficiency on a Single GPU! Hugging Face TRL and RapidFire AI's Super-Parallel Revolution
Single-GPU Super-Parallel Fine-TuningHugging Face TRLHyperparameter OptimizationLLM Post-TrainingRapidFire AI
...
The Significance of Gemini 3: AI Has Surpassed the 'Hallucination Phase', Approaching Humans, 'Human-Machine Collaboration' Will Shift from 'Humans Correcting AI' to 'Humans Guiding AI Work'
Gemini 3AI AgentsAI EvolutionEthan MollickHuman-AI Collaboration
...
Coding Tests Crush Humans! Claude Opus 4.5 Midnight Surprise, AI Coding Enters 'Superhuman Era'
Claude Opus 4.5AI CodingToolchain UpgradesAgentsSWE-Bench
...
[In-Depth] Ilya Sutskever's Selected Paper: The Platonic Representation Hypothesis
Platonic Representation HypothesisAI DevelopmentPlato's Cave AllegoryRepresentation LearningDeep Learning
...
NVIDIA Plans to Invest Up to $1 Billion in This AI Coding Startup! Led by Copilot Tech Guru, Valued at Nearly $30 Billion in Two Years
NVIDIA Investment in Poolside AIAI Coding AssistantReinforcement Learning for CodingFunding RoundsJason Warner
...
AI Beats Human PhDs for the First Time, Top Conference Papers Turn into Code Instantly! HKU Post-90s Open-Source Project Explodes to 8k Stars
DeepCodePaper ReproductionHKU TeamOpen SourceAI Code Generation
...
Claude Launches Skills Feature and Agent Skills Development Guide
Claude SkillsAgent SkillsAI AgentsAnthropicDevelopment Guide
...
Recently, The New Yorker Has an Article with a Scary Title: Will Artificial Intelligence Trap You in a "Permanent Underclass"?
Artificial IntelligencePermanent UnderclassCyber FeudalismFuture of WorkAI Unemployment
...
Oxford University Proposes H1 "Bootstrapping" Framework, Enabling Models to Self-Evolve with Emergent Abilities!
Long-Horizon ReasoningBootstrappingLLMsProblem ChainingCurriculum Reinforcement Learning
...
Under $8,000! Sina Weibo's 1.5B Small Model Surpasses Near-Trillion Parameter Models
Artificial IntelligenceMachine LearningCost EfficiencyModel TrainingLarge Language Models
...
RL Remembers More Firmly, SFT More Forgetful? Princeton Danqi Chen's Team Rewrites Post-Training Cognition
LLM Post-TrainingSFT vs RLMemory RetentionOn-Policy DataCatastrophic Forgetting
...
Understanding neural networks through sparse circuits
Sparse CircuitsNeural NetworksAI SafetyMechanistic InterpretabilityInterpretability
...
Xiaohongshu Proposes DeepEyesV2: From "Visual Thinking" to "Tool Collaboration", Exploring New Dimensions in Multimodal Intelligence
DeepEyesV2Multimodal ReasoningRealX-BenchReinforcement LearningTool Collaboration
...
Microsoft CEO Nadella: This Industrial Revolution Starts with the "AI Superfactory"
AI SuperfactoryMicrosoftAI InfrastructureData CentersNadella
...
Microsoft Proposes GAD Framework: Open-Source Models Can Directly Distill Black-Box GPT-5
GAD FrameworkBlack-Box DistillationLarge Language ModelsReinforcement LearningGenerative Adversarial Distillation
...
Making LLMs Work Like a Company: Microsoft Turns 'Concurrent Thinking' into a Protocol, Higher Accuracy and 28% Reduction in Critical Path Latency
Artificial IntelligenceLarge Language ModelsAsyncThinkMicrosoft ResearchConcurrent Thinking
...
Letting CoT "Evolve" with the Environment: AgileThinker Achieves "Thinking While Doing" | Latest from Tsinghua
AI AgentsReal-Time ReasoningReactive AgentsPlanning AgentsDual-System Theory
...
AI Cracks 18th-Century "Mystery" Ledger in Seconds! Google's New Model Blind Test Goes Viral
Artificial IntelligenceHandwritten Text RecognitionGoogle GeminiHistorical DocumentsMachine Learning
...
Reinforcement Learning + Large Model Memory: Mem-α, Enabling Agents to "Learn How to Remember" for the First Time
Reinforcement LearningLarge Language ModelsArtificial IntelligenceAI AgentsMemory Management
...