@iwiwi
Our inference-time scaling method AB-MCTS (NeurIPS’25 Spotlight✨) is already powering multiple real-world projects. By combining strong domain knowledge with inference-time scaling, we’re tackling important, hard problems that vanilla LLMs struggled with. 🚀 "Wider or Deeper? Scaling LLM Inference-Time Compute with Adaptive Branching Tree Search" 📄Paper: https://t.co/tz20wmJaDp 🌐Blog: https://t.co/xrVFegEeKd 💻Code: https://t.co/MFRJJNxtwy