\"Be Concise\" Halves Tokens, Lifts Accuracy by 16 Points

Featured

• DARE: Aligning LLM Agents with the R Statistical Ecosystem via Distribution-Aware Retrieval score 10
  入选 HF Daily Papers; HF 热度: 42 upvotes (+4); 有代码实现; 关键词(4): lightweight, retrieval-augmented, code generation, open-source
• Locality-Attending Vision Transformer score 10
  入选 HF Daily Papers; HF 热度: 6 upvotes (+2); 有代码实现; 顶会接收: ICLR
• Interactive Benchmarks score 9
  入选 HF Daily Papers; HF 热度: 14 upvotes (+3); 有代码实现; 关键词(1): reasoning
• MASQuant: Modality-Aware Smoothing Quantization for Multimodal Large Language Models score 9
  入选 HF Daily Papers; HF 热度: 8 upvotes (+2); 关键词(2): quantization, post-training; 顶会接收: CVPR
• RoboPocket: Improve Robot Policies Instantly with Your Phone score 8
  入选 HF Daily Papers; HF 热度: 30 upvotes (+4); 关键词(2): scaling, finetuning
• RealWonder: Real-Time Physical Action-Conditioned Video Generation score 8
  入选 HF Daily Papers; HF 热度: 6 upvotes (+2); 有代码实现; 关键词(1): real-time
• On-Policy Self-Distillation for Reasoning Compression score 8
  入选 HF Daily Papers; HF 热度: 5 upvotes (+2); 有代码实现; 关键词(3): compression, distillation, reasoning
• Towards Multimodal Lifelong Understanding: A Dataset and Agentic Baseline score 7
  入选 HF Daily Papers; HF 热度: 2 upvotes (+1); 有代码实现; 关键词(1): agentic
• Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling score 6
  入选 HF Daily Papers; HF 热度: 9 upvotes (+2); 关键词(5): scaling, post-training, pre-training, MoE, leaderboard
• KARL: Knowledge Agents via Reinforcement Learning score 5
  入选 HF Daily Papers; HF 热度: 4 upvotes (+1); 关键词(6): latency, post-training, agentic, tool use, reasoning

Also Worth Noting

• FOZO: Forward-Only Zeroth-Order Prompt Optimization for Test-Time Adaptation score 4
  关键词(1): deployment; 顶会接收: CVPR
• Toward Real-world Infrared Image Super-Resolution: A Unified Autoregressive Framework and Benchmark Dataset score 4
  关键词(1): quantization; 顶会接收: CVPR
• EchoGuard: An Agentic Framework with Knowledge-Graph Memory for Detecting Manipulative Communication in Longitudinal Dialogue score 4
  机构: Microsoft; 关键词(1): agentic
• DeformTrace: A Deformable State Space Model with Relay Tokens for Temporal Forgery Localization score 4
  关键词(2): state space, reasoning; 顶会接收: AAAI
• FedAFD: Multimodal Federated Learning via Adversarial Fusion and Distillation score 4
  关键词(1): distillation; 顶会接收: CVPR
• Think, Then Verify: A Hypothesis-Verification Multi-Agent Framework for Long Video Understanding score 4
  关键词(1): reasoning; 顶会接收: CVPR
• CoIn3D: Revisiting Configuration-Invariant Multi-Camera 3D Object Detection score 4
  关键词(1): deployment; 顶会接收: CVPR
• MoRe: Motion-aware Feed-forward 4D Reconstruction Transformer score 4
  关键词(2): real-time, fine-tune; 顶会接收: CVPR
• Sparse-BitNet: 1.58-bit LLMs are Naturally Friendly to Semi-Structured Sparsity score 4
  机构: Peking University; 关键词(2): quantization, pretraining
• Wiki-R1: Incentivizing Multimodal Reasoning for Knowledge-based VQA via Data and Sampling Curriculum score 4
  关键词(3): post-training, pretraining, reasoning; 顶会接收: ICLR
• Layer by layer, module by module: Choose both for optimal OOD probing of ViT score 4
  关键词(1): pretraining; 顶会接收: ICLR
• Latent Wasserstein Adversarial Imitation Learning score 4
  关键词(2): pre-training, state space; 顶会接收: ICLR
• NaiLIA: Multimodal Nail Design Retrieval Based on Dense Intent Descriptions and Palette Queries score 4
  关键词(1): vision-language; 顶会接收: CVPR
• Beyond Scattered Acceptance: Fast and Coherent Inference for DLMs via Longest Stable Prefixes score 4
  关键词(2): code generation, reasoning; 顶会接收: ICLR
• Cheap Thrills: Effective Amortized Optimization Using Inexpensive Labels score 4
  机构: Carnegie Mellon; 关键词(1): pretraining