Category: Large Language Models

• When Async Agentic RL Meets 'Amnesia for Old Policies': Rethinking Off-Policy Correction
• HIT KaLM-Reranker-V1: An Efficient RAG Retrieval Booster, Deep Dive into High-Speed Reranking
• Large Models Finally Stop Swearing! Toxic Subword Pruning ToxPrune: Dual Defense at Pre-training and Inference
• Community Submission | Bailin Ling & Ring 2.6 Technical Report Released: Efficient Trillion-Parameter Models for Real-World Agent Workflows
• ACL 2026 | Why Does SFT Always Fail to Learn? Not Every SFT Failure Needs More Epochs! Five "Surgical Scalpels" to Fix SFT
• The End of Code Review: Coding Agents Supersede Human Inspection
• Anthropic's First Mythic-Class Claude 5 Officially Unleashed
• Anthropic Participates in New Paper: Why Do Larger Models Learn More? The Answer Lies in Scaling
• 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
• Open Source Update | Are LLMs Still Stuck with 'Goldfish Memory'? New Benchmark RHELM Tests the Ceiling of 'Real Long-Term Memory'
• Latest Discovery: AI Large Models Know When They're Being Evaluated
• Google and Cornell's New Research: The Next Step for LLMs Is Learning How to 'Sleep' Well
• Meta Skill Has Just Arrived
• Models Are Too Fond of Cheating! Cursor Reveals the Inside Story of Composer 2's Reinforcement Learning: Models Can Detect 'Fake Environments', and Floating-Point Non-Determinism Is a Fatal Flaw in RL Training
• Stop Handwriting Skills! Microsoft’s Latest Research: Train Your Skills Like Neural Networks
• Unbelievable… FaceWall Had AI Write a Training Framework, and It Trained the Strongest 1B Model by Itself: MiniCPM5-1B
• On 520, Meet China's New 'Model King' Qwen3.7-Max!
• Burning Too Many Tokens Using Claude Code on Large Codebases? This Open-Source Tool Cuts Tool Calls by 92%
• 35B-Parameter Science Model Rivals Trillion-Parameter Giants: 'Intern' Science Model Intern-S2-Preview Open-Sourced
• WWW'26 | A New Paradigm for Cross-Task Adaptive Multi-Agent Collaboration
• Math Major, in Crisis! Fields Medalist Tests ChatGPT 5.5 Pro, Achieves Thesis-Level Result in 17 Minutes
• Zero Index, Zero Embedding, Pure Grep: DCI Does Deep Research Directly on Raw Corpora
• Attention Is All You Need Author Returns: Can a 99% Sparse Transformer Be Even Faster?
• The Ghost of Markov: From Predicting the Next Word to Predicting the Next Action
• Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies in LLM Reinforcement Learning