ずんだもんのHugging Faceニュース

【本日の論文】 1. SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning https://huggingface.co/papers/2602.13515 2. Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents https://huggingface.co/papers/2602.16855 3. Unified Latents (UL): How to train your latents https://huggingface.co/papers/2602.17270 4. Frontier AI Risk Management Framework in...

【本日の論文】 1. SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning https://huggingface.co/papers/2602.13515 2. Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents https://huggingface.co/papers/2602.16855 3. Unified Latents (UL): How to train your latents https://huggingface.co/papers/2602.17270 4. Frontier AI Risk Management Framework in...

【本日の論文】 1. SpargeAttention2: Trainable Sparse Attention via Hybrid Top-k+Top-p Masking and Distillation Fine-Tuning https://huggingface.co/papers/2602.13515 2. Unified Latents (UL): How to train your latents https://huggingface.co/papers/2602.17270 3. Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents https://huggingface.co/papers/2602.16855 4. "What Are You Doing?": Effects of Interm...

【本日の論文】 1. SLA2: Sparse-Linear Attention with Learnable Routing and QAT https://huggingface.co/papers/2602.12675 2. RynnBrain: Open Embodied Foundation Models https://huggingface.co/papers/2602.14979 3. Learning Humanoid End-Effector Control for Open-Vocabulary Visual Loco-Manipulation https://huggingface.co/papers/2602.16705 4. CADEvolve: Creating Realistic CAD via Program Evolution https:...

【本日の論文】 1. Sanity Checks for Sparse Autoencoders: Do SAEs Beat Random Baselines? https://huggingface.co/papers/2602.14111 2. SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks https://huggingface.co/papers/2602.12670 3. GLM-5: from Vibe Coding to Agentic Engineering https://huggingface.co/papers/2602.15763 4. Does Socialization Emerge in AI Agent Society? A Case Study...

【本日の論文】 1. Experiential Reinforcement Learning https://huggingface.co/papers/2602.13949 2. DeepImageSearch: Benchmarking Multimodal Agents for Context-Aware Image Retrieval in Visual Histories https://huggingface.co/papers/2602.10809 3. REDSearcher: A Scalable and Cost-Efficient Framework for Long-Horizon Search Agents https://huggingface.co/papers/2602.14234 4. STATe-of-Thoughts: Structure...

【本日の論文】 1. Less is Enough: Synthesizing Diverse Data in Feature Space of LLMs https://huggingface.co/papers/2602.10388 2. SQuTR: A Robustness Benchmark for Spoken Query to Text Retrieval under Acoustic Noise https://huggingface.co/papers/2602.12783 3. MedXIAOHE: A Comprehensive Recipe for Building Medical MLLMs https://huggingface.co/papers/2602.12705 4. Zooming without Zooming: Region-to-I...

【本日の論文】 1. CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding https://huggingface.co/papers/2602.01785 2. AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration https://huggingface.co/papers/2602.03786 3. No Global Plan in Chain-of-Thought: Uncover the Latent Planning Horizon of LLMs https://huggingface.co/papers/2602.02103 4. MARS: Modular Agent with...

【本日の論文】 1. Green-VLA: Staged Vision-Language-Action Model for Generalist Robots https://huggingface.co/papers/2602.00919 2. Kimi K2.5: Visual Agentic Intelligence https://huggingface.co/papers/2602.02276 3. Vision-DeepResearch: Incentivizing DeepResearch Capability in Multimodal Large Language Models https://huggingface.co/papers/2601.22060 4. Vision-DeepResearch Benchmark: Rethinking Visua...

【本日の論文】 1. Idea2Story: An Automated Pipeline for Transforming Research Concepts into Complete Scientific Narratives https://huggingface.co/papers/2601.20833 2. Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models https://huggingface.co/papers/2601.20354 3. Scaling Embeddings Outperforms Scaling Experts in Language Models https://huggingface.co/papers/2601.21204...

【本日の論文】 1. Idea2Story: An Automated Pipeline for Transforming Research Concepts into Complete Scientific Narratives https://huggingface.co/papers/2601.20833 2. Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models https://huggingface.co/papers/2601.20354 3. Scaling Embeddings Outperforms Scaling Experts in Language Models https://huggingface.co/papers/2601.21204...

【本日の論文】 1. Idea2Story: An Automated Pipeline for Transforming Research Concepts into Complete Scientific Narratives https://huggingface.co/papers/2601.20833 2. Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models https://huggingface.co/papers/2601.20354 3. Scaling Embeddings Outperforms Scaling Experts in Language Models https://huggingface.co/papers/2601.21204...

【本日の論文】 1. Harder Is Better: Boosting Mathematical Reasoning via Difficulty-Aware GRPO and Multi-Aspect Question Reformulation https://huggingface.co/papers/2601.20614 2. Advancing Open-source World Models https://huggingface.co/papers/2601.20540 3. Innovator-VL: A Multimodal Large Language Model for Scientific Discovery https://huggingface.co/papers/2601.19325 4. DeepSeek-OCR 2: Visual Cau...

【本日の論文】 1. AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security https://huggingface.co/papers/2601.18491 2. AdaReasoner: Dynamic Tool Orchestration for Iterative Visual Reasoning https://huggingface.co/papers/2601.18631 3. A Pragmatic VLA Foundation Model https://huggingface.co/papers/2601.18692 4. Visual Generation Unlocks Human-Like Reasoning through Multimodal Worl...

【本日の論文】 1. Can LLMs Clean Up Your Mess? A Survey of Application-Ready Data Preparation with LLMs https://huggingface.co/papers/2601.17058 2. daVinci-Dev: Agent-native Mid-training for Software Engineering https://huggingface.co/papers/2601.18418 3. The Script is All You Need: An Agentic Framework for Long-Horizon Dialogue-to-Cinematic Video Generation https://huggingface.co/papers/2601.1773...

【本日の論文】 1. LongCat-Flash-Thinking-2601 Technical Report https://huggingface.co/papers/2601.16725 2. SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents https://huggingface.co/papers/2601.16746 3. TwinBrainVLA: Unleashing the Potential of Generalist VLMs for Embodied Tasks via Asymmetric Mixture-of-Transformers https://huggingface.co/papers/2601.14133 4. VisGym: Diverse, Customizable...

【本日の論文】 1. EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience https://huggingface.co/papers/2601.15876 2. HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding https://huggingface.co/papers/2601.14724 3. LLM-in-Sandbox Elicits General Agentic Intelligence https://huggingface.co/papers/2601.16206 4. The Flexibility Trap: Why Arb...

【本日の論文】 1. EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience https://huggingface.co/papers/2601.15876 2. HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding https://huggingface.co/papers/2601.14724 3. The Flexibility Trap: Why Arbitrary Order Limits Reasoning Potential in Diffusion Language Models https://huggingface.co/pape...

【本日の論文】 1. EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience https://huggingface.co/papers/2601.15876 2. The Flexibility Trap: Why Arbitrary Order Limits Reasoning Potential in Diffusion Language Models https://huggingface.co/papers/2601.15165 3. HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding https://huggingface.co/pape...

【本日の論文】 1. Agentic Reasoning for Large Language Models https://huggingface.co/papers/2601.12538 2. MMDeepResearch-Bench: A Benchmark for Multimodal Deep Research Agents https://huggingface.co/papers/2601.12346 3. Rethinking Video Generation Model for the Embodied World https://huggingface.co/papers/2601.15282 4. Paper2Rebuttal: A Multi-Agent Framework for Transparent Author Response Assista...

【本日の論文】 1. Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization https://huggingface.co/papers/2601.12993 2. Advances and Frontiers of LLM-based Issue Resolution in Software Engineering: A Comprehensive Survey https://huggingface.co/papers/2601.11655 3. Toward Efficient Agents: Memory, Tool learning, and Planning https://huggingface.co/papers/2601.14192 4. Omn...

【本日の論文】 1. ABC-Bench: Benchmarking Agentic Backend Coding in Real-World Development https://huggingface.co/papers/2601.11077 2. Multiplex Thinking: Reasoning via Token-wise Branch-and-Merge https://huggingface.co/papers/2601.08808 3. Medical SAM3: A Foundation Model for Universal Prompt-Driven Medical Image Segmentation https://huggingface.co/papers/2601.10880 4. NAACL: Noise-AwAre Verbal C...