List your company
ResearchFormerly WizWand

Class conditional image generation on ImageNet 64x64 (train test)

1.02FIDUni-Instruct
0.62083.31546.018.7046Feb 18, 2021Nov 4, 2021Jul 22, 2022Apr 8, 2023Dec 24, 2023Sep 9, 2024May 27, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Uni-InstructFamily=Few Step, NFE=1, Divergence=FKL, Training Setting=resume and Longer Training2025.051.02——
SiD2AFamily=Few Step, NFE=12025.051.1——
f-distillFamily=Few Step, NFE=12025.051.16——
RINFamily=Diffusion, NFE=10002025.051.23——
MMDFamily=Few Step, NFE=82025.051.24——
DMD2Family=Few Step, NFE=1, Training=longer training2025.051.28——
Uni-InstructFamily=Few Step, NFE=1, Divergence=JKL, Training Setting=from scratch2025.051.28——
Uni-InstructFamily=Few Step, NFE=1, Divergence=FKL, Training Setting=from scratch2025.051.34——
SIDAFamily=Few Step, NFE=12025.051.35——
Uni-InstructFamily=Few Step, NFE=1, Divergence=RKL, Training Setting=from scratch2025.051.35——
VP-EDMFamily=Teacher, NFE=5112025.051.36——
SiDFamily=Few Step, NFE=12025.051.71——
CTMFamily=Few Step, NFE=12025.051.92——
G-distillFamily=Few Step, NFE=82025.052.05——
ADMFamily=Diffusion, NFE=2502025.052.07——
EMD-16Family=Few Step, NFE=12025.052.2——
ECTFamily=Consistency, NFE=12025.052.49——
DiT-L/2Family=Diffusion, NFE=2502025.052.91——
Improved Diffusion (large)Parameters=270M, Training iterations=250K, Sampling steps=250, Objective=L_hybrid2021.022.928271
iCT-deepFamily=Consistency, NFE=12025.053.25——
iCT-deepFamily=Few Step, NFE=12025.053.25——
ICTFamily=Consistency, NFE=12025.054.02——
BigGAN-deepParameters=100M, Training iterations=125K, Truncation=None2021.024.068659
U-ViTFamily=Diffusion, NFE=502025.054.26——
Diff-InstructFamily=Few Step, NFE=12025.055.57——
Improved Diffusion (small)Parameters=100M, Training iterations=1.7M, Sampling steps=250, Objective=L_hybrid2021.026.927772
DFNOFamily=Few Step, NFE=12025.057.83——
PDFamily=Few Step, NFE=22025.058.95——
PIDFamily=Few Step, NFE=12025.059.49——
DDPMFamily=Diffusion, NFE=2502025.0511——