Review

[논문리뷰] Expertise need not monopolize: Action-Specialized Mixture of Experts for Vision-Language-Action Learning

Sijia Gu이 arXiv에 게시한 'Expertise need not monopolize: Action-Specialized Mixture of Experts for Vision-Language-Action Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language-Action (VLA)#Mixture of Experts (MoE)#Robotic Manipulation #Expert Specialization #Decoupled Routing #Load Balancing #Transfer Learning

2025년 10월 17일

[논문리뷰] Efficient Parallel Samplers for Recurrent-Depth Models and Their Connection to Diffusion Language Models

arXiv에 게시된 'Efficient Parallel Samplers for Recurrent-Depth Models and Their Connection to Diffusion Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Recurrent-Depth Models #Diffusion Forcing #Parallel Sampling #LLM Inference Acceleration #Transformer Architectures #Generative AI #Latent Space Diffusion

2025년 10월 17일

[논문리뷰] DialectGen: Benchmarking and Improving Dialect Robustness in Multimodal Generation

arXiv에 게시된 'DialectGen: Benchmarking and Improving Dialect Robustness in Multimodal Generation' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal Generation #Dialect Robustness #Text-to-Image #Text-to-Video #Benchmarking #Diffusion Models #Text Encoder Tuning #Low-Resource Dialects

2025년 10월 17일

[논문리뷰] COIG-Writer: A High-Quality Dataset for Chinese Creative Writing with Thought Processes

arXiv에 게시된 'COIG-Writer: A High-Quality Dataset for Chinese Creative Writing with Thought Processes' 논문에 대한 자세한 리뷰입니다.

#Review #Chinese Creative Writing #Process Supervision #LLM Training #Dataset Creation #Cross-Lingual Transfer #Narrative Logic #Linguistic Expression #Type-Token Ratio

2025년 10월 17일

[논문리뷰] BitNet Distillation

arXiv에 게시된 'BitNet Distillation' 논문에 대한 자세한 리뷰입니다.

#Review #Low-bit Quantization #LLM Compression #Knowledge Distillation #Ternary Weights #Inference Optimization #Memory Efficiency #SubLN #Continual Pre-training

2025년 10월 17일

[논문리뷰] Beyond One World: Benchmarking Super Heros in Role-Playing Across Multiversal Contexts

arXiv에 게시된 'Beyond One World: Benchmarking Super Heros in Role-Playing Across Multiversal Contexts' 논문에 대한 자세한 리뷰입니다.

#Review #Role-playing LLMs #Multiversal Consistency #Character Benchmarking #Moral Dilemmas #Canon Events #Reasoning-Acting Alignment #Chain-of-Thought #Superheroes

2025년 10월 17일

[논문리뷰] Beyond Correctness: Evaluating Subjective Writing Preferences Across Cultures

arXiv에 게시된 'Beyond Correctness: Evaluating Subjective Writing Preferences Across Cultures' 논문에 대한 자세한 리뷰입니다.

#Review #Subjective Preference Learning #Writing Evaluation #Reward Models #RLHF #Cross-Cultural AI #Generative Models #Language Model Judges #Genre Instability

2025년 10월 17일

[논문리뷰] Attention Is All You Need for KV Cache in Diffusion LLMs

arXiv에 게시된 'Attention Is All You Need for KV Cache in Diffusion LLMs' 논문에 대한 자세한 리뷰입니다.

#Review #Diffusion LLMs #KV Cache #Adaptive Caching #Inference Optimization #Attention Mechanism #Latency Reduction #Generative AI

2025년 10월 17일

[논문리뷰] Agentic Entropy-Balanced Policy Optimization

arXiv에 게시된 'Agentic Entropy-Balanced Policy Optimization' 논문에 대한 자세한 리뷰입니다.

#Review #Agentic Reinforcement Learning #Web Agents #Tool Learning #Entropy Balancing #Policy Optimization #Rollout Strategy #Large Language Models

2025년 10월 17일

[논문리뷰] AI for Service: Proactive Assistance with AI Glasses

arXiv에 게시된 'AI for Service: Proactive Assistance with AI Glasses' 논문에 대한 자세한 리뷰입니다.

#Review #AI for Service #Proactive AI #AI Glasses #Multi-agent System #Human-AI Interaction #Context-aware AI #Wearable AI

2025년 10월 17일

[논문리뷰] X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model

Xirui Kang이 arXiv에 게시한 'X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language-Action (VLA) Models #Soft Prompts #Transformer #Cross-Embodiment #Robotics #Pretraining #Domain Adaptation #Flow Matching

2025년 10월 16일

[논문리뷰] Universal Image Restoration Pre-training via Masked Degradation Classification

arXiv에 게시된 'Universal Image Restoration Pre-training via Masked Degradation Classification' 논문에 대한 자세한 리뷰입니다.

#Review #Universal Image Restoration #Pre-training #Masked Image Modeling #Degradation Classification #Deep Learning #Computer Vision #Self-supervised Learning #Low-level Vision

2025년 10월 16일

[논문리뷰] UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity MoE

arXiv에 게시된 'UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity MoE' 논문에 대한 자세한 리뷰입니다.

#Review #Mixture of Experts #Speech Generation #Music Generation #Multimodal AI #Dynamic Routing #Training Curriculum #Data Imbalance #Audio Synthesis

2025년 10월 16일

[논문리뷰] UniME-V2: MLLM-as-a-Judge for Universal Multimodal Embedding Learning

Ziyong Feng이 arXiv에 게시한 'UniME-V2: MLLM-as-a-Judge for Universal Multimodal Embedding Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal Embeddings #MLLM-as-a-Judge #Hard Negative Mining #Semantic Alignment #Representation Learning #Reranking #Contrastive Learning

2025년 10월 16일

[논문리뷰] Uni-MMMU: A Massive Multi-discipline Multimodal Unified Benchmark

arXiv에 게시된 'Uni-MMMU: A Massive Multi-discipline Multimodal Unified Benchmark' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal AI #Unified Models #Benchmark #Generation #Understanding #Reasoning #Evaluation #Cross-modal Synergy

2025년 10월 16일

[논문리뷰] Trace Anything: Representing Any Video in 4D via Trajectory Fields

arXiv에 게시된 'Trace Anything: Representing Any Video in 4D via Trajectory Fields' 논문에 대한 자세한 리뷰입니다.

#Review #4D Video Representation #Trajectory Fields #Neural Networks #Spatio-temporal Modeling #3D Point Tracking #Motion Forecasting #Computer Vision #B-splines

2025년 10월 16일

[논문리뷰] The Role of Computing Resources in Publishing Foundation Model Research

Zhenwen Liang이 arXiv에 게시한 'The Role of Computing Resources in Publishing Foundation Model Research' 논문에 대한 자세한 리뷰입니다.

#Review #Foundation Models #Computing Resources #GPU Disparity #AI Research #Publication Bias #Resource Allocation #Research Transparency

2025년 10월 16일

[논문리뷰] The Art of Scaling Reinforcement Learning Compute for LLMs

arXiv에 게시된 'The Art of Scaling Reinforcement Learning Compute for LLMs' 논문에 대한 자세한 리뷰입니다.

#Review #Reinforcement Learning #LLMs #Scaling Laws #Compute Efficiency #Predictability #Sigmoidal Curves #ScaleRL #Off-Policy RL

2025년 10월 16일

[논문리뷰] Stronger Together: On-Policy Reinforcement Learning for Collaborative LLMs

Hao Zhang이 arXiv에 게시한 'Stronger Together: On-Policy Reinforcement Learning for Collaborative LLMs' 논문에 대한 자세한 리뷰입니다.

#Review #Large Language Models (LLMs)#Reinforcement Learning (RL)#Multi-Agent Systems (MAS)#On-Policy RL #Collaborative AI #Agentic LLMs #Group-based Optimization

2025년 10월 16일

[논문리뷰] Revisiting Model Interpolation for Efficient Reasoning

arXiv에 게시된 'Revisiting Model Interpolation for Efficient Reasoning' 논문에 대한 자세한 리뷰입니다.

#Review #Model Interpolation #Efficient Reasoning #Large Language Models #Chain-of-Thought #Model Merging #Performance Dynamics #Ablation Study

2025년 10월 16일