Image Generation On Stl 10

评估指标

FID

Inception score

评测结果

各个模型在此基准测试上的表现结果

模型名称	FID	Inception score	Paper Title	Repository
TransGAN	18.28	10.43	TransGAN: Two Pure Transformers Can Make One Strong GAN, and That Can Scale Up
Dist-GAN	36.19	-	Dist-GAN: An Improved GAN using Distance Constraints
SN-GAN	40.1	9.10	Spectral Normalization for Generative Adversarial Networks
WaveDiff	12.93	-	Wavelet Diffusion Models are fast and scalable Image Generators
MMD-PMish-NAS	11.61	11.79	Enhancing GANs with MMD Neural Architecture Search, PMish Activation Function, and Adaptive Rank Decomposition
DC-VAE	41.9	8.1	Dual Contradistinctive Generative Autoencoder	-
RDUOT	11.5	-	A High-Quality Robust Diffusion Framework for Corrupted Dataset
AutoGAN	31.01	9.16	AutoGAN: Neural Architecture Search for Generative Adversarial Networks
PeerGAN	51.37	-	DuelGAN: A Duel Between Two Discriminators Stabilizes the GAN Training	-
RDGAN	13.07	-	A High-Quality Robust Diffusion Framework for Corrupted Dataset
UNCSN++ (RVE) + ST	7.71	13.43	Soft Truncation: A Universal Training Technique of Score-based Diffusion Model for High Precision Score Estimation
aw-SN-GAN	34.72	9.61	Adaptive Weighted Discriminator for Training Generative Adversarial Networks
MMD-AdversarialNAS	12.91	11.6	Enhancing GAN Performance through Neural Architecture Search and Tensor Decomposition
D2GAN	-	7.98	Dual Discriminator Generative Adversarial Nets
SNGAN-DCD (Pixel)	22.25	9.25	Discriminator Contrastive Divergence: Semi-Amortized Generative Modeling by Exploring Energy of the Discriminator
QSNGAN	59.611	4.987	Quaternion Generative Adversarial Networks
MMD-AdversarialNAS (Compressed Small)	14.84	11.66	Enhancing GAN Performance through Neural Architecture Search and Tensor Decomposition
EAGAN (G)	23.34	10.02	EAGAN: Efficient Two-stage Evolutionary Architecture Search for GANs
MMD-PMish-NAS (Compressed)	13.07	11.85	Enhancing GANs with MMD Neural Architecture Search, PMish Activation Function, and Adaptive Rank Decomposition
SNGAN-DCD (Latent)	17.68	9.33	Discriminator Contrastive Divergence: Semi-Amortized Generative Modeling by Exploring Energy of the Discriminator

0 of 31 row(s) selected.