Stop When Reasoning Converges, Save 26% of Tokens

Featured

• Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models score 9
  入选 HF Daily Papers; HF 热度: 19 upvotes (+3); 有代码实现; 关键词(5): lightweight, latency, code generation, reasoning, vision-language
• LiteFrame: Efficient Vision Encoders Unlock Frame Scaling in Video LLMs score 9
  入选 HF Daily Papers; HF 热度: 19 upvotes (+3); 有代码实现; 关键词(3): scaling, distillation, latency

Also Worth Noting

• Active Budget Allocation for Efficient Scaling Law Estimation via Surrogate-Guided Pruning score 4
  关键词(2): scaling, pruning; 顶会接收: ICML
• A Hybrid Gaussian Process Regression Framework for Stable Volatility-Covariance Estimation: Evidence from Global Equity Indices score 3
  机构: Google
• UNR-Explainer: Counterfactual Explanations for Unsupervised Node Representation Learning Models score 3
  顶会接收: ICLR