Thinking in Loops: Scaling Visual ARC with Looped Transformers
Looped (weight-tied) Transformers for Visual ARC—scale reasoning with “thinking time”.
Enhancing Diffusion-based Restoration Models via Difficulty-Adaptive Reinforcement Learning with IQA Reward
Difficulty-adaptive RL for diffusion restoration with IQA reward.
Go with Your Gut: Scaling Confidence for Autoregressive Image Generation
Confidence scaling for better autoregressive image generation.
ThinkVid: Benchmarking Visual Reasoning in Video Generative Models
Benchmarking visual reasoning in video generation.
CoRe-GRPO: Consensus-driven and Region-focused RL for Human-Centric Generation in Lightweight Diffusion Models
Region-focused RL for human-centric diffusion generation.
AlignVid: Training-Free Attention Scaling for Semantic Fidelity in Text-Guided Image-to-Video Generation
Training-free attention scaling for text-to-video semantic fidelity.
Temporal Regularization Makes Your Video Generator Stronger
Temporal regularization improves video generator quality and stability.
LightGen: Efficient Image Generation through Knowledge Distillation and Direct Preference Optimization
Knowledge distillation + DPO for efficient image generation.
VideoGen-of-Thought: Step-by-step Generating Multi-shot Video with Minimal Manual Intervention
Step-by-step multi-shot video generation with minimal manual intervention.