Unconditional Image Generation on ImageNet (train val) benchmark leaderboard · SOTA2 Research

Unconditional Image Generation on ImageNet (train val)

8.72FIDDiT-XL/2-256

Evaluation Results

Method	Links
DiT-XL/2-2562024.11		8.72	69	76	70	84	58.15	0.33
DiT-XL/2-5122024.11		9.54	70	73	72	82	55.17	0.34
MAR-B-2562024.11		10.36	72	71	72	75	55.78	0.38
MAR-L-2562024.11		10.36	72	71	73	75	55.95	0.38
MAR-H-2562024.11		10.36	72	71	74	76	55.82	0.38
ADM-2562024.11		30.3	57	73	41	40	36.18	0.2
LDM-2562024.11		37.39	61	68	45	28	30.83	0.16
EDM-2-XS-5122024.11		75.02	42	63	25	13	26.95	0.09
EDM-2-M-5122024.11		107.45	36	61	19	9	22.19	0.06
EDM-2-L-5122024.11		118.87	23	51	11	6	22.89	0.03
EDM-2-S-5122024.11		122.48	33	42	16	7	21.34	0.04
EDM-2-XXL-5122024.11		124.29	33	60	17	7	21.52	0.05
EDM-2-XL-5122024.11		141.74	25	45	12	6	20.04	0.03