List your company

Text-to-Image Generation on MS-COCO 256x256 (val)

5.4FIDU-ViT-S/2 (Deep)
4.196412.320720.44528.5693Dec 20, 2021Aug 6, 2022Mar 24, 2023Nov 9, 2023Jun 26, 2024Feb 11, 2025Sep 29, 2025

Evaluation results

MethodLinks
U-ViT-S/2 (Deep)Type=Diffusion, NFE=50, Resolution=256x2562025.095.4——
U-ViT-S/2Type=Diffusion, NFE=50, Resolution=256x2562025.095.95——
ImagenEvaluation Protocol=Zero-shot, Guidance Weight (64x64)=1.35, Guidance Weight (super-resolution)=82022.057.27——
U-NetType=Diffusion, NFE=50, Resolution=256x2562025.097.32——
Make-A-SceneZero-shot=false2022.047.55——
Make-A-SceneEvaluation Protocol=Trained on MS-COCO2022.057.55——
EasyGenEvaluation Protocol=Fine-tuned on MS-COCO, Model Type=MM-Model2023.107.68——
EasyGen w/o adapterEvaluation Protocol=Fine-tuned on MS-COCO, Model Type=MM-Model2023.107.89——
MM-GATType=GAN, NFE=1, CLIP image encoder=false, Resolution=256x2562025.097.98——
LAFITEZero-shot=false2021.128.12——
LAFITEZero-shot=false2022.048.12——
LAFITEEvaluation Protocol=Trained on MS-COCO2022.058.12——
UniDiffuserEvaluation Protocol=Fine-tuned on MS-COCO, Model Type=Diffusion Model2023.108.12——
LAFITEType=GAN, NFE=1, CLIP image encoder=true, Resolution=256x2562025.098.12——
FridoType=Diffusion, NFE=200, Resolution=256x2562025.098.97——
EasyGenEvaluation Protocol=Zero-Shot, Model Type=MM-Model2023.109.16——
XMC-GANZero-shot=false2021.129.33——
XMC-GANZero-shot=false2022.049.33——
XMC-GANEvaluation Protocol=Trained on MS-COCO2022.059.33——
XMC-GANType=GAN, NFE=1, Resolution=256x2562025.099.33——
EasyGen w/o adapterEvaluation Protocol=Zero-Shot, Model Type=MM-Model2023.109.52——
UniDiffuserEvaluation Protocol=Zero-Shot, Model Type=Diffusion Model2023.109.71——
SD v1.5Evaluation Protocol=Zero-Shot, Model Type=Diffusion Model2023.109.93——
unCLIPPrior type=Diffusion prior, Zero-shot=true, Filtering=false, Guidance scale=1.252022.0410.39——
DALL-E 2Evaluation Protocol=Zero-shot2022.0510.39——
unCLIPPrior type=AR prior, Zero-shot=true, Filtering=false, Guidance scale=1.252022.0410.63——
unCLIPPrior type=Diffusion prior, Zero-shot=true, Filtering=true, Guidance scale=1.252022.0410.87——
unCLIPPrior type=AR prior, Zero-shot=true, Filtering=true, Guidance scale=1.252022.0411.08——
SDEvaluation Protocol=Zero-Shot, Model Type=Diffusion Model2023.1011.21——
NExT-GPTEvaluation Protocol=Zero-Shot, Model Type=MM-Model2023.1011.28——
EmuEvaluation Protocol=Zero-Shot, Model Type=MM-Model2023.1011.66——
Make-A-SceneNparams=4B, Classifier-free guidance (c.f.g.)=s=52021.1211.84——
Make-A-SceneZero-shot=true, Filtering=true2022.0411.84——
GLIDEZero-shot=true, Guidance Scale=1.52021.1212.24——
GLIDENparams=6B, Sampling steps=277 DDIM steps, Classifier-free guidance (c.f.g.)=s=32021.1212.24——
GLIDEZero-shot=true, Filtering=false2022.0412.24——
GLIDEEvaluation Protocol=Zero-shot2022.0512.24——
LDM-KL-8-GNparams=1.45B, Sampling steps=250 DDIM steps, Classifier-free guidance (c.f.g.)=s=1.52021.1212.63——
GLIDEZero-shot=true, Filtering=true2022.0412.89——
DALL-EEvaluation Protocol=Zero-shot2022.0517.89——
DF-GANType=GAN, NFE=1, Resolution=256x2562025.0919.32——
VQ-DiffusionType=Discrete diffusion, NFE=100, Resolution=256x2562025.0919.75——
DM-GAN + CLZero-shot=false2021.1220.79——
DM-GAN + CLZero-shot=false2022.0420.79——
DM-GAN + CLEvaluation Protocol=Trained on MS-COCO2022.0520.79——
DF-GANZero-shot=false2021.1221.42——
DF-GANZero-shot=false2022.0421.42——
DF-GANEvaluation Protocol=Trained on MS-COCO2022.0521.42——
LDM-KL-8Nparams=1.45B, Sampling steps=250 DDIM steps2021.1223.31——
LAFITEZero-shot=true2021.1226.94——
LAFITENparams=75M2021.1226.9426.02—
LAFITEZero-shot=true, Filtering=false2022.0426.94——
LAFITEEvaluation Protocol=Zero-shot2022.0526.94——
CogViewNparams=4B, Self-ranking=rejection rate 0.0172021.1227.118.2—
DALL-EZero-shot=true2021.1228——
DALL-EZero-shot=true, Filtering=false2022.0428——
DM-GANZero-shot=false2021.1232.64——
DM-GANZero-shot=false2022.0432.64——
DM-GANEvaluation Protocol=Trained on MS-COCO2022.0532.64——
DM-GANType=GAN, NFE=1, Resolution=256x2562025.0932.64——
AttnGANZero-shot=false2021.1235.49——
AttnGANZero-shot=false2022.0435.49——
AttnGANEvaluation Protocol=Trained on MS-COCO2022.0535.49——
AttnGANType=GAN, NFE=1, Resolution=256x2562025.0935.49——
CogViewEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——27.1
CogView2Evaluation Protocol=Zero-shot, Resolution=256x2562022.10——24
DALL-EEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——27.5
DALL-E 2Evaluation Protocol=Zero-shot, Resolution=256x2562022.10——10.39
ERNIE-ViLGEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——14.7
ERNIE-ViLG 2.0Evaluation Protocol=Zero-shot, Resolution=256x256, classifier-free guidance scale=2.1, parameters=24B, denoising experts=102022.10——6.75
GLIDEEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——12.24
ImagenEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——7.27
LAFITEEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——26.94
LDMEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——12.61
Make-A-SceneEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——11.84
PartiEvaluation Protocol=Zero-shot, Resolution=256x2562022.10——7.23