「想一想」能翻模型的记忆库

重点关注

• Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs score 11
  机构: Google；入选 HF Daily Papers；HF 热度: 55 upvotes (+4)；关键词(2): code generation, reasoning
• OpenClaw-RL: Train Any Agent Simply by Talking score 10
  入选 HF Daily Papers；HF 热度: 67 upvotes (+4)；有代码实现；关键词(2): distillation, agentic
• MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data score 10
  入选 HF Daily Papers；HF 热度: 41 upvotes (+4)；有代码实现；关键词(2): GRPO, reasoning
• InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing score 10
  入选 HF Daily Papers；HF 热度: 35 upvotes (+4)；有代码实现；关键词(2): lightweight, reasoning
• Stepping VLMs onto the Court: Benchmarking Spatial Intelligence in Sports score 10
  入选 HF Daily Papers；HF 热度: 24 upvotes (+4)；有代码实现；关键词(4): fine-tuning, reasoning, vision-language, open-source
• MA-EgoQA: Question Answering over Egocentric Videos from Multiple Embodied Agents score 10
  入选 HF Daily Papers；HF 热度: 24 upvotes (+4)；有代码实现；关键词(2): reasoning, embodied
• The Reasoning Trap -- Logical Reasoning as a Mechanistic Pathway to Situational Awareness score 9
  入选 HF Daily Papers；HF 热度: 5 upvotes (+2)；关键词(2): deployment, reasoning；顶会接收: ICLR
• Reading, Not Thinking: Understanding and Bridging the Modality Gap When Text Becomes Pixels in Multimodal LLMs score 8
  入选 HF Daily Papers；HF 热度: 23 upvotes (+4)；关键词(2): distillation, reasoning
• MiniAppBench: Evaluating the Shift from Text to Interactive HTML Responses in LLM-Powered Assistants score 9
  入选 HF Daily Papers；HF 热度: 11 upvotes (+3)；有代码实现；关键词(2): agentic, code generation
• Streaming Autoregressive Video Generation via Diagonal Distillation score 8
  入选 HF Daily Papers；HF 热度: 5 upvotes (+2)；有代码实现；关键词(3): distillation, latency, real-time

也值得关注

• Code-Space Response Oracles: Generating Interpretable Multi-Agent Policies with Large Language Models score 4
  入选 HF Daily Papers；关键词(1): code generation
• Time, Identity and Consciousness in Language Model Agents score 4
  关键词(1): tool use；顶会接收: AAAI
• From Days to Minutes: An Autonomous AI Agent Achieves Reliable Clinical Triage in Remote Patient Monitoring score 4
  机构: Stanford；关键词(1): reasoning
• A Text-Native Interface for Generative Video Authoring score 4
  入选 HF Daily Papers；关键词(1): deployment
• Chain of Event-Centric Causal Thought for Physically Plausible Video Generation score 4
  关键词(2): reasoning, vision-language；顶会接收: CVPR
• Evoking User Memory: Personalizing LLM via Recollection-Familiarity Adaptive Retrieval score 4
  关键词(2): latency, reasoning；顶会接收: ICLR
• OddGridBench: Exposing the Lack of Fine-Grained Visual Discrepancy Sensitivity in Multimodal Large Language Models score 4
  关键词(2): GRPO, open-source；顶会接收: CVPR
• TaSR-RAG: Taxonomy-guided Structured Reasoning for Retrieval-Augmented Generation score 4
  机构: Alibaba；关键词(4): lightweight, retrieval-augmented, RAG, reasoning
• From Flow to One Step: Real-Time Multi-Modal Trajectory Policies via Implicit Maximum Likelihood Estimation-based Distribution Distillation score 4
  机构: Google；关键词(3): distillation, latency, real-time
• CIGPose: Causal Intervention Graph Neural Network for Whole-Body Pose Estimation score 4
  关键词(1): reasoning；顶会接收: CVPR
• Context-Nav: Context-Driven Exploration and Viewpoint-Aware 3D Spatial Reasoning for Instance Navigation score 4
  关键词(2): fine-tuning, reasoning；顶会接收: CVPR
• You Didn't Have to Say It like That: Subliminal Learning from Faithful Paraphrases score 4
  机构: Peking University；关键词(1): synthetic data
• More than the Sum: Panorama-Language Models for Adverse Omni-Scenes score 4
  关键词(2): reasoning, vision-language；顶会接收: CVPR
• BinaryAttention: One-Bit QK-Attention for Vision and Diffusion Transformers score 4
  关键词(2): quantization, distillation；顶会接收: CVPR
• ES-dLLM: Efficient Inference for Diffusion Large Language Models by Early-Skipping score 4
  关键词(1): throughput；顶会接收: ICLR