#Transformer Architecture

53개의 포스트

[논문리뷰] Communication-Inspired Tokenization for Structured Image Representations

arXiv에 게시된 'Communication-Inspired Tokenization for Structured Image Representations' 논문에 대한 자세한 리뷰입니다.

#Review #Image Tokenization #Structured Representation #Attentive Encoding #Flow Matching #Semantic Alignment #Compositional Generalization #Transformer Architecture

2026년 2월 24일

[논문리뷰] Large Causal Models for Temporal Causal Discovery

Dario Simionato이 arXiv에 게시한 'Large Causal Models for Temporal Causal Discovery' 논문에 대한 자세한 리뷰입니다.

#Review #Causal Discovery #Temporal Models #Foundation Models #Transformer Architecture #Zero-shot Learning #Time-series Data #Scalability #Multi-dataset Pretraining

2026년 2월 23일

[논문리뷰] Arcee Trinity Large Technical Report

arXiv에 게시된 'Arcee Trinity Large Technical Report' 논문에 대한 자세한 리뷰입니다.

#Review #Mixture-of-Experts #Sparse LLM #Training Stability #Load Balancing #MoE #Transformer Architecture #Context Extension #Muon Optimizer

2026년 2월 19일

[논문리뷰] MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models

arXiv에 게시된 'MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models' 논문에 대한 자세한 리뷰입니다.

#Review #Audio Tokenizer #Transformer Architecture #End-to-End Learning #Residual Vector Quantization #Speech Synthesis #Audio Foundation Models #Scalability #Autoregressive Models

2026년 2월 12일

[논문리뷰] HySparse: A Hybrid Sparse Attention Architecture with Oracle Token Selection and KV Cache Sharing

arXiv에 게시된 'HySparse: A Hybrid Sparse Attention Architecture with Oracle Token Selection and KV Cache Sharing' 논문에 대한 자세한 리뷰입니다.

#Review #Sparse Attention #KV Cache Sharing #Hybrid Attention #Long-Context LLMs #Memory Optimization #Token Selection #Transformer Architecture

2026년 2월 4일

[논문리뷰] DeepSeek-OCR 2: Visual Causal Flow

arXiv에 게시된 'DeepSeek-OCR 2: Visual Causal Flow' 논문에 대한 자세한 리뷰입니다.

#Review #OCR #Vision-Language Model #Causal Reasoning #Transformer Architecture #Attention Mechanism #Document Understanding #DeepEncoder

2026년 1월 28일

[논문리뷰] Post-LayerNorm Is Back: Stable, ExpressivE, and Deep

arXiv에 게시된 'Post-LayerNorm Is Back: Stable, ExpressivE, and Deep' 논문에 대한 자세한 리뷰입니다.

#Review #Transformer Architecture #Layer Normalization #Depth Scaling #Training Stability #Large Language Models #Gradient Flow #Highway Networks #Post-LayerNorm

2026년 1월 27일

[논문리뷰] SkyReels-V3 Technique Report

arXiv에 게시된 'SkyReels-V3 Technique Report' 논문에 대한 자세한 리뷰입니다.

#Review #Video Generation #Multimodal AI #Diffusion Models #Transformer Architecture #Reference-guided Generation #Video-to-Video #Audio-driven Animation #Temporal Consistency

2026년 1월 26일

[논문리뷰] Parallel Latent Reasoning for Sequential Recommendation

Yuning Jiang이 arXiv에 게시한 'Parallel Latent Reasoning for Sequential Recommendation' 논문에 대한 자세한 리뷰입니다.

#Review #Sequential Recommendation #Latent Reasoning #Parallel Processing #Computational Scaling #Mixture of Experts #Contrastive Learning #Transformer Architecture

2026년 1월 6일

[논문리뷰] Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers

arXiv에 게시된 'Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers' 논문에 대한 자세한 리뷰입니다.

#Review #Language Models #Transformer Architecture #Canon Layers #Synthetic Pretraining #Reasoning Depth #Linear Attention #State-Space Models #NoPE

2025년 12월 21일

[논문리뷰] REGLUE Your Latents with Global and Local Semantics for Entangled Diffusion

Giorgos Sfikas이 arXiv에 게시한 'REGLUE Your Latents with Global and Local Semantics for Entangled Diffusion' 논문에 대한 자세한 리뷰입니다.

#Review #Latent Diffusion Models #Vision Foundation Models #Semantic Compression #Global-Local Semantics #Image Generation #Representation Entanglement #Transformer Architecture

2025년 12월 18일

[논문리뷰] Stronger Normalization-Free Transformers

Zhuang Liu이 arXiv에 게시한 'Stronger Normalization-Free Transformers' 논문에 대한 자세한 리뷰입니다.

#Review #Normalization-Free Transformers #Point-wise Functions #Error Function #Deep Learning #Transformer Architecture #Generalization #Normalization Layers

2025년 12월 11일

[논문리뷰] MoCapAnything: Unified 3D Motion Capture for Arbitrary Skeletons from Monocular Videos

Qi Wang이 arXiv에 게시한 'MoCapAnything: Unified 3D Motion Capture for Arbitrary Skeletons from Monocular Videos' 논문에 대한 자세한 리뷰입니다.

#Review #3D Motion Capture #Monocular Video #Arbitrary Skeletons #Motion Retargeting #Deep Learning #Inverse Kinematics #Transformer Architecture #Category-Agnostic

2025년 12월 11일

[논문리뷰] Efficiently Reconstructing Dynamic Scenes One D4RT at a Time

arXiv에 게시된 'Efficiently Reconstructing Dynamic Scenes One D4RT at a Time' 논문에 대한 자세한 리뷰입니다.

#Review #Dynamic Scene Reconstruction #4D Reconstruction #Point Tracking #Transformer Architecture #Feedforward Model #Query-based Inference #Computer Vision #Geometric Consistency

2025년 12월 9일

[논문리뷰] FMA-Net++: Motion- and Exposure-Aware Real-World Joint Video Super-Resolution and Deblurring

Munchurl Kim이 arXiv에 게시한 'FMA-Net++: Motion- and Exposure-Aware Real-World Joint Video Super-Resolution and Deblurring' 논문에 대한 자세한 리뷰입니다.

#Review #Video Super-Resolution #Video Deblurring #Joint Restoration #Exposure-Aware #Motion Compensation #Transformer Architecture #Dynamic Filtering #Real-World Degradations

2025년 12월 4일

[논문리뷰] Mixture of Horizons in Action Chunking

Zelong Sun이 arXiv에 게시한 'Mixture of Horizons in Action Chunking' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language-Action Models #Action Chunking #Robotic Manipulation #Multi-horizon Planning #Transformer Architecture #Gated Fusion #Dynamic Inference

2025년 12월 2일

[논문리뷰] Does Hearing Help Seeing? Investigating Audio-Video Joint Denoising for Video Generation

arXiv에 게시된 'Does Hearing Help Seeing? Investigating Audio-Video Joint Denoising for Video Generation' 논문에 대한 자세한 리뷰입니다.

#Review #Video Generation #Audio-Video Multimodal #Joint Denoising #Diffusion Models #Transformer Architecture #World Models #Physical Commonsense #Multimodal Training

2025년 12월 2일

[논문리뷰] Adversarial Flow Models

arXiv에 게시된 'Adversarial Flow Models' 논문에 대한 자세한 리뷰입니다.

#Review #Generative Models #Adversarial Flow Models #GANs #Flow Matching #Optimal Transport #Single-step Generation #Image Generation #Transformer Architecture

2025년 11월 30일

[논문리뷰] Terminal Velocity Matching

Jiaming Song이 arXiv에 게시한 'Terminal Velocity Matching' 논문에 대한 자세한 리뷰입니다.

#Review #Generative Models #Flow Matching #Diffusion Models #One-Step Generation #Few-Step Generation #Wasserstein Distance #Transformer Architecture #Lipschitz Continuity

2025년 11월 26일

[논문리뷰] ReDirector: Creating Any-Length Video Retakes with Rotary Camera Encoding

arXiv에 게시된 'ReDirector: Creating Any-Length Video Retakes with Rotary Camera Encoding' 논문에 대한 자세한 리뷰입니다.

#Review #Video Retake Generation #Camera Control #Rotary Position Embedding (RoPE)#Rotary Camera Encoding (RoCE)#Geometric Consistency #Video Generative Models #Transformer Architecture #Multi-view Synthesis

2025년 11월 25일

[논문리뷰] MergeDNA: Context-aware Genome Modeling with Dynamic Tokenization through Token Merging

arXiv에 게시된 'MergeDNA: Context-aware Genome Modeling with Dynamic Tokenization through Token Merging' 논문에 대한 자세한 리뷰입니다.

#Review #Genome Modeling #Dynamic Tokenization #Token Merging #Context-aware Learning #DNA Foundation Models #Transformer Architecture #Multi-omics

2025년 11월 23일

[논문리뷰] Mixture of States: Routing Token-Level Dynamics for Multimodal Generation

arXiv에 게시된 'Mixture of States: Routing Token-Level Dynamics for Multimodal Generation' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal Diffusion #Mixture of States (MoS)#Token-Level Routing #Dynamic Conditional Fusion #Text-to-Image Generation #Image Editing #Transformer Architecture

2025년 11월 19일

[논문리뷰] A Brain Wave Encodes a Thousand Tokens: Modeling Inter-Cortical Neural Interactions for Effective EEG-based Emotion Recognition

G. Maragatham이 arXiv에 게시한 'A Brain Wave Encodes a Thousand Tokens: Modeling Inter-Cortical Neural Interactions for Effective EEG-based Emotion Recognition' 논문에 대한 자세한 리뷰입니다.

#Review #EEG #Emotion Recognition #Transformer Architecture #Inter-Cortical Neural Interactions #Multi-Head Attention #Brain-Computer Interface #Affective Computing

2025년 11월 18일

[논문리뷰] Depth Anything 3: Recovering the Visual Space from Any Views

arXiv에 게시된 'Depth Anything 3: Recovering the Visual Space from Any Views' 논문에 대한 자세한 리뷰입니다.

#Review #Depth Estimation #Multi-view Geometry #Transformer Architecture #Teacher-Student Learning #Pose Estimation #3D Reconstruction #Novel View Synthesis #Visual Space Recovery

2025년 11월 13일

[논문리뷰] EVTAR: End-to-End Try on with Additional Unpaired Visual Reference

arXiv에 게시된 'EVTAR: End-to-End Try on with Additional Unpaired Visual Reference' 논문에 대한 자세한 리뷰입니다.

#Review #Virtual Try-on #Diffusion Models #End-to-End Learning #Reference Images #Unpaired Data #Flow Matching #Transformer Architecture #Generative AI

2025년 11월 9일

[논문리뷰] Forget BIT, It is All about TOKEN: Towards Semantic Information Theory for LLMs

Bo Bai이 arXiv에 게시한 'Forget BIT, It is All about TOKEN: Towards Semantic Information Theory for LLMs' 논문에 대한 자세한 리뷰입니다.

#Review #Semantic Information Theory #Large Language Models #Directed Information #Rate-Distortion Function #Granger Causality #Token Embedding #Transformer Architecture #Variational Inference

2025년 11월 9일

[논문리뷰] The End of Manual Decoding: Towards Truly End-to-End Language Models

arXiv에 게시된 'The End of Manual Decoding: Towards Truly End-to-End Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Large Language Models (LLMs)#End-to-End Generation #Dynamic Decoding #Hyperparameter Optimization #Stochastic Sampling #Instruction Following #Transformer Architecture

2025년 10월 31일

[논문리뷰] Scaling Latent Reasoning via Looped Language Models

arXiv에 게시된 'Scaling Latent Reasoning via Looped Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Looped Language Models #Latent Reasoning #Parameter Efficiency #Adaptive Computation #Pre-training Scaling #Knowledge Manipulation #Early Exit Mechanisms #Transformer Architecture

2025년 10월 30일

[논문리뷰] DyPE: Dynamic Position Extrapolation for Ultra High Resolution Diffusion

arXiv에 게시된 'DyPE: Dynamic Position Extrapolation for Ultra High Resolution Diffusion' 논문에 대한 자세한 리뷰입니다.

#Review #Diffusion Models #Transformer Architecture #Positional Encoding #High-Resolution Image Generation #Extrapolation #Dynamic Adaptation #Training-Free

2025년 10월 24일

[논문리뷰] Attention Sinks in Diffusion Language Models

Simone Scardapane이 arXiv에 게시한 'Attention Sinks in Diffusion Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Diffusion Language Models #Attention Sinks #Transformer Architecture #Masked Language Modeling #Bidirectional Attention #Generative Models #Robustness #Dynamic Attention

2025년 10월 23일

[논문리뷰] From Pixels to Words -- Towards Native Vision-Language Primitives at Scale

arXiv에 게시된 'From Pixels to Words -- Towards Native Vision-Language Primitives at Scale' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language Models #Native VLMs #Early Fusion #Multimodal Learning #Transformer Architecture #Rotary Position Embeddings #Pixel-Word Alignment #End-to-End Training

2025년 10월 17일

[논문리뷰] Direct Multi-Token Decoding

Xifeng Yan이 arXiv에 게시한 'Direct Multi-Token Decoding' 논문에 대한 자세한 리뷰입니다.

#Review #LLM Inference #Multi-token Decoding #Transformer Architecture #Layer Specialization #Cyclical Refilling #Inference Speedup #Model Scaling

2025년 10월 16일

[논문리뷰] Towards Scalable and Consistent 3D Editing

Pan Zhou이 arXiv에 게시한 'Towards Scalable and Consistent 3D Editing' 논문에 대한 자세한 리뷰입니다.

#Review #3D Editing #Generative Models #Transformer Architecture #Dataset Generation #Multimodal Learning #Conditional Generation #Image-to-3D

2025년 10월 10일

[논문리뷰] Online Generic Event Boundary Detection

Jonghyun Choi이 arXiv에 게시한 'Online Generic Event Boundary Detection' 논문에 대한 자세한 리뷰입니다.

#Review #Online Video Analysis #Event Boundary Detection #Event Segmentation Theory #Real-time AI #Anomaly Detection #Transformer Architecture

2025년 10월 9일

[논문리뷰] Native Hybrid Attention for Efficient Sequence Modeling

Yu Cheng이 arXiv에 게시한 'Native Hybrid Attention for Efficient Sequence Modeling' 논문에 대한 자세한 리뷰입니다.

#Review #Sequence Modeling #Hybrid Attention #Transformer Architecture #Linear Attention #Sliding Window Attention #Long Context #Large Language Models (LLMs)#Efficiency

2025년 10월 9일

[논문리뷰] Hybrid Architectures for Language Models: Systematic Analysis and Design Insights

arXiv에 게시된 'Hybrid Architectures for Language Models: Systematic Analysis and Design Insights' 논문에 대한 자세한 리뷰입니다.

#Review #Hybrid LLM #Transformer Architecture #Mamba #State Space Models (SSM)#Computational Efficiency #Long-Context #Language Model Architectures #Scaling Laws

2025년 10월 7일

[논문리뷰] Improving GUI Grounding with Explicit Position-to-Coordinate Mapping

Spandana Gella이 arXiv에 게시한 'Improving GUI Grounding with Explicit Position-to-Coordinate Mapping' 논문에 대한 자세한 리뷰입니다.

#Review #GUI Grounding #Vision-Language Models #Positional Embedding #UI Automation #Coordinate Prediction #Resolution Generalization #Transformer Architecture

2025년 10월 6일

[논문리뷰] Boolean Satisfiability via Imitation Learning

Xiangyu Xu이 arXiv에 게시한 'Boolean Satisfiability via Imitation Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Boolean Satisfiability #Imitation Learning #CDCL Solvers #Branching Policy #KeyTrace #Transformer Architecture #Perceiver AR

2025년 10월 2일

[논문리뷰] jina-reranker-v3: Last but Not Late Interaction for Document Reranking

arXiv에 게시된 'jina-reranker-v3: Last but Not Late Interaction for Document Reranking' 논문에 대한 자세한 리뷰입니다.

#Review #Document Reranking #Last but Not Late Interaction #Multilingual #Transformer Architecture #Cross-Encoder #InfoNCE Loss #Contextual Embedding #Qwen3

2025년 10월 1일

[논문리뷰] Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training

arXiv에 게시된 'Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training' 논문에 대한 자세한 리뷰입니다.

#Review #Mechanistic Interpretability #Attention Heads #Post-Training #Supervised Fine-Tuning (SFT)#Reinforcement Learning (RL)#Circuit Analysis #Reasoning Models #Transformer Architecture

2025년 10월 1일

[논문리뷰] The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain

arXiv에 게시된 'The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain' 논문에 대한 자세한 리뷰입니다.

#Review #Large Language Models #Brain-Inspired AI #Graph Neural Networks #Hebbian Learning #Scale-Free Networks #Model Interpretability #Transformer Architecture

2025년 10월 1일

[논문리뷰] X-Streamer: Unified Human World Modeling with Audiovisual Interaction

Guoxian Song이 arXiv에 게시한 'X-Streamer: Unified Human World Modeling with Audiovisual Interaction' 논문에 대한 자세한 리뷰입니다.

#Review #Digital Human #Multimodal AI #Real-time Streaming #Video Generation #Diffusion Models #Transformer Architecture #Audiovisual Synchronization #World Modeling

2025년 9월 29일

[논문리뷰] AToken: A Unified Tokenizer for Vision

Mingze Xu이 arXiv에 게시한 'AToken: A Unified Tokenizer for Vision' 논문에 대한 자세한 리뷰입니다.

#Review #Unified Visual Tokenizer #Multimodal AI #Transformer Architecture #4D Representation #Adversarial-free Training #Reconstruction #Semantic Understanding #Generative Models

2025년 9월 19일

[논문리뷰] InfGen: A Resolution-Agnostic Paradigm for Scalable Image Synthesis

Song Guo이 arXiv에 게시한 'InfGen: A Resolution-Agnostic Paradigm for Scalable Image Synthesis' 논문에 대한 자세한 리뷰입니다.

#Review #Image Synthesis #Resolution-Agnostic #Diffusion Models #Latent Space #VAE Decoder #High-Resolution Image Generation #Generative AI #Transformer Architecture

2025년 9월 15일

[논문리뷰] Drawing2CAD: Sequence-to-Sequence Learning for CAD Generation from Vector Drawings

Meie Fang이 arXiv에 게시한 'Drawing2CAD: Sequence-to-Sequence Learning for CAD Generation from Vector Drawings' 논문에 대한 자세한 리뷰입니다.

#Review #CAD Generation #Vector Graphics #Sequence-to-Sequence Learning #Transformer Architecture #Engineering Drawings #Multi-modal Learning #Soft Target Loss #Dual Decoder

2025년 9월 5일

[논문리뷰] MotionFlux: Efficient Text-Guided Motion Generation through Rectified Flow Matching and Preference Alignment

An-An Liu이 arXiv에 게시한 'MotionFlux: Efficient Text-Guided Motion Generation through Rectified Flow Matching and Preference Alignment' 논문에 대한 자세한 리뷰입니다.

#Review #Text-Guided Motion Generation #Rectified Flow Matching #Preference Alignment #Human Motion Synthesis #Real-time AI #Transformer Architecture #Self-supervised Learning

2025년 8월 28일

[논문리뷰] Wan-S2V: Audio-Driven Cinematic Video Generation

Chaonan Ji이 arXiv에 게시한 'Wan-S2V: Audio-Driven Cinematic Video Generation' 논문에 대한 자세한 리뷰입니다.

#Review #Audio-Driven Video Generation #Cinematic Video #Diffusion Models #Transformer Architecture #Long Video Consistency #Human Animation #Multimodal Control #Data Curation

2025년 8월 27일

[논문리뷰] UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning

Ran Guo이 arXiv에 게시한 'UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Memory Networks #Mixture of Experts (MoE)#Long-Context Learning #Sparse Models #Transformer Architecture #LLMs #Efficient Inference

2025년 8월 27일

[논문리뷰] Advances in Speech Separation: Techniques, Challenges, and Future Trends

Zhuo Chen이 arXiv에 게시한 'Advances in Speech Separation: Techniques, Challenges, and Future Trends' 논문에 대한 자세한 리뷰입니다.

#Review #Speech Separation #Deep Neural Networks #Cocktail Party Problem #Transformer Architecture #Unsupervised Learning #Supervised Learning #Evaluation Metrics #Datasets

2025년 8월 20일

[논문리뷰] NextStep-1: Toward Autoregressive Image Generation with Continuous Tokens at Scale

Quan Sun이 arXiv에 게시한 'NextStep-1: Toward Autoregressive Image Generation with Continuous Tokens at Scale' 논문에 대한 자세한 리뷰입니다.

#Review #Autoregressive Models #Text-to-Image Generation #Continuous Latent Tokens #Flow Matching #Image Editing #Multimodal Learning #Transformer Architecture

2025년 8월 15일

[논문리뷰] Exploitation Is All You Need... for Exploration

Jesse Roberts이 arXiv에 게시한 'Exploitation Is All You Need... for Exploration' 논문에 대한 자세한 리뷰입니다.

#Review #Reinforcement Learning #Exploration-Exploitation #Meta-RL #Transformer Architecture #Emergent Behavior #Multi-Armed Bandits #Gridworlds #Pseudo-Thompson Sampling

2025년 8월 5일

[논문리뷰] PixNerd: Pixel Neural Field Diffusion

Limin Wang이 arXiv에 게시한 'PixNerd: Pixel Neural Field Diffusion' 논문에 대한 자세한 리뷰입니다.

#Review #Diffusion Models #Neural Fields #Pixel Space #Generative Models #Image Synthesis #Transformer Architecture #End-to-End Learning

2025년 8월 4일

[논문리뷰] iLRM: An Iterative Large 3D Reconstruction Model

Abdelrahman Mohamed이 arXiv에 게시한 'iLRM: An Iterative Large 3D Reconstruction Model' 논문에 대한 자세한 리뷰입니다.

#Review #3D Reconstruction #Gaussian Splatting #Iterative Refinement #Transformer Architecture #Multi-view Learning #Scalability #Feed-forward Models

2025년 8월 2일