ImageNet 512×512

'ImageNet 512×512' is a benchmark dataset used to evaluate the quality of image synthesis models, specifically measuring their performance in generating high-resolution images.

Papers using ImageNet 512×512 (11)

Simplifying, Stabilizing and Scaling Continuous-Time Consistency Models2024 · 1 citesTerminal Velocity Matching2025DC-AE 1.5: Accelerating Diffusion Model Convergence with Structured Latent Space2025Latent Denoising Makes Good Tokenizers2025Align Your Flow: Scaling Continuous-Time Flow Map Distillation2025Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator2025Diffusion Models Beat GANs on Image Synthesis2021 · 2,179 citesScalable Diffusion Models with Transformers2022 · 42 citesSiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers2024 · 35 citesFast Training of Diffusion Models with Masked Transformers2023 · 8 citesDiffFit: Unlocking Transferability of Large Diffusion Models via Simple Parameter-Efficient Fine-Tuning2023 · 5 cites