Iris-3B: Going Beyond the Latent with Pixel-Space Diffusion Training, Conversion and Fine-Tuning

Benchmark Model Rank Results
monocular-depth-estimation-on-kitti-relative-log-alignedIris-3B#3AbsRel: 0.083delta1: 0.927
monocular-depth-estimation-on-scannet-relative-log-alignedIris-3B#4AbsRel: 0.069delta1: 0.939
image-super-resolution-on-div2k-val-real-esrgan-4Iris-3B#6LPIPS: 0.292SSIM: 0.481MUSIQ: 68.01PSNR: 20.68
text-to-image-generation-on-oneig-enIris-3B#13OneIG Overall Score: 0.54
text-to-image-generation-on-longtext-benchIris-3B#18Accuracy: 85.7
text-to-image-generation-on-dpg-benchIris-3B#38DPG-Bench Score: 86.52
text-to-image-generation-on-genevalIris-3B#105Overall: 0.798