ImageNet-256

ImageNet-256 is a dataset used to evaluate the performance of generative models, containing a subset of images from the larger ImageNet dataset, specifically resized to 256x256 pixels.

Papers using ImageNet-256 (12)

CAR-Flow: Condition-Aware Reparameterization Aligns Source and Target for Better Flow Matching2025Unlearning for One-Step Generative Models via Unbalanced Optimal Transport2026PixelGen: Pixel Diffusion Beats Latent Diffusion with Perceptual Loss2026Boosting Latent Diffusion Models via Disentangled Representation Alignment2026PixelDiT: Pixel Diffusion Transformers for Image Generation2025Adversarial Flow Models2025There is No VAE: End-to-End Pixel-Space Generative Modeling via Self-Supervised Pre-training2025Soft-Di[M]O: Improving One-Step Discrete Image Generation with Soft Embeddings2025Scalable GANs with Transformers2025Diffusion Models without Classifier-free Guidance2025Improving Diffusion Model Efficiency Through Patching2022 · 5 citesAccelerating Guided Diffusion Sampling with Splitting Numerical Methods2023 · 3 cites