AI 연구 논문 데일리

Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach

Goku: Flow Based Video Generative Foundation Models

VideoRoPE: What Makes for Good Video Rotary Position Embedding?

Fast Video Generation with Sliding Tile Attention

QuEST: Stable Training of LLMs with 1-Bit Weights and Activations

AuraFusion360: Augmented Unseen Region Alignment for Reference-based 360° Unbounded Scene Inpainting

FlashVideo:Flowing Fidelity to Detail for Efficient High-Resolution Video Generation

DuoGuard: A Two-Player RL-Driven Framework for Multilingual LLM Guardrails

Agency Is Frame-Dependent

Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Models

Generating Symbolic World Models via Test-time Scaling of Large Language Models

CMoE: Fast Carving of Mixture-of-Experts for Efficient LLM Inference

On-device Sora: Enabling Diffusion-Based Text-to-Video Generation for Mobile Devices

CodeSteer: Symbolic-Augmented Language Models via Code/Text Guidance

No Task Left Behind: Isotropic Model Merging with Common and Task-Specific Subspaces

Scaling Laws in Patchification: An Image Is Worth 50,176 Tokens And More

QLIP: Text-Aligned Visual Tokenization Unifies Auto-Regressive Multimodal Understanding and Generation

Linear Correlation in LM's Compositional Generalization and Hallucination

Lost in Time: Clock and Calendar Understanding Challenges in Multimodal LLMs

ARR: Question Answering with Large Language Models via Analyzing, Retrieving, and Reasoning

Value-Based Deep RL Scales Predictably

YINYANG-ALIGN: Benchmarking Contradictory Objectives and Proposing Multi-Objective Optimization based DPO for Text-to-Image Alignment

Adaptive Semantic Prompt Caching with VectorQ

Continuous 3D Perception Model with Persistent State

MEETING DELEGATE: Benchmarking LLMs on Attending Meetings on Our Behalf

SPARC: Subspace-Aware Prompt Adaptation for Robust Continual Learning in LLMs

Intelligent Sensing-to-Action for Robust Autonomy at the Edge: Opportunities and Challenges