DreamSim
A perceptual image similarity metric used to measure visual differences between images. It is commonly applied to assess the diversity of generated image batches in text-to-image tasks.
Timeline (last 90 days)
On the Pick-a-Pic dataset using DreamSim as the diversity metric, with FLUX.1-dev as the base model and HPSv3 as the reward model, SatisDive improved the worst candidate reward by up to 0.43 compared to FK steering.
DreamSim is a deep learning-based perceptual similarity metric trained to align with human judgments of image similarity, and it is closer to subjective visual experience than LPIPS or FID.
All Facts (2)
On the Pick-a-Pic dataset using DreamSim as the diversity metric, with FLUX.1-dev as the base model and HPSv3 as the reward model, SatisDive improved the worst candidate reward by up to 0.43 compared to FK steering.
50%UnverifiedDreamSim is a deep learning-based perceptual similarity metric trained to align with human judgments of image similarity, and it is closer to subjective visual experience than LPIPS or FID.
50%