Image Generation On Celeba 64X64

Metriken

FID

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Vergleichstabelle

Modellname	FID
high-resolution-deep-convolutional-generative	8.44
Modell 2	1.77
densely-connected-normalizing-flows	-
escaping-from-collapsing-modes-in-a	34.136
consistency-regularization-for-variational	-
efficient-vdvae-less-is-more	-
Modell 7	6.03
soft-diffusion-score-matching-for-general	1.85
learning-stackable-and-skippable-lego-bricks	2.09
blackout-diffusion-generative-diffusion	3.22
diffusion-gan-training-gans-with-diffusion	1.69
enhancing-gans-with-mmd-neural-architecture	1.92
compensation-sampling-for-improved	2.11
enhancing-gans-with-mmd-neural-architecture	2.03
score-matching-model-for-unbounded-data-score-1	2.9
optimal-budgeted-rejection-sampling-for	3.74
scire-solver-efficient-sampling-of-diffusion	2.02
refining-generative-process-with	1.34
clr-gan-improving-gans-stability-and-quality	13.63
flow-contrastive-estimation-of-energy-based	12.21
feature-alignment-for-approximated	128.35
efficient-generative-adversarial-networks	1.81
input-perturbation-reduces-exposure-bias-in	1.27
peergan-generative-adversarial-networks-with	13.95
score-matching-model-for-unbounded-data-score-1	-
maximum-likelihood-training-of-implicit	2.54
diffusevae-efficient-controllable-and-high	3.97
continuous-time-functional-diffusion	35
score-matching-model-for-unbounded-data-score-1	1.9
continuous-time-functional-diffusion	11
maximum-likelihood-training-of-implicit	1.75
Modell 32	22.45
transgan-two-transformers-can-make-one-strong	12.23
compensation-sampling-for-improved	1.38
pseudo-numerical-methods-for-diffusion-models-1	2.71
learning-energy-based-models-by-diffusion-1	5.98
ncp-vae-variational-autoencoders-with-noise-1	5.25
probabilistic-auto-encoder	49.2
class-continuous-conditional-generative	5.6