Text-to-Image Generation on ImageNet (FID, Precision, Recall, Coverage, Density, Vendi, CLIP) benchmark leaderboard · SOTA2 Research

Text-to-Image Generation on ImageNet (FID, Precision, Recall, Coverage, Density, Vendi, CLIP)

17.13FIDOriginal

Evaluation Results

Method	Links
Original2026.06		17.13	81.44	51.06	80.32	0.9311	2.05	28.61
DAVE Random2026.06		17.57	85.91	54.22	83.71	1.0114	2.5	29.39
CADS2026.06		17.91	78.56	56.98	79.36	0.8386	2.09	29.07
DAVE2026.06		20.74	80.9	64.89	78.95	0.781	2.33	28.97
Original2026.06		21	82.2	40.64	77.26	0.9475	2.09	28.52
Original2026.06		22.23	86.04	25.89	77.57	1.0071	1.71	29.52
SPARKE2026.06		22.27	86.86	31.36	40.04	1.2289	1.77	30.16
DAVE Random2026.06		22.41	74.19	49.11	69.9	0.7935	2.15	28.79
Original2026.06		22.44	79.86	28.46	71.82	0.8963	1.8	28.27
SPARKE2026.06		22.65	82.98	35.61	39.35	1.1112	2.04	29.21
Original2026.06		24.01	82.19	31.49	75.55	0.9371	1.88	29.02
DAVE Random2026.06		24.36	69.27	60.32	72.95	0.6869	2.44	28.18
DAVE2026.06		25.16	75.88	52.54	64.35	0.6422	2.2	28.85
DAVE2026.06		25.31	71.12	51.02	69.63	0.7189	2.26	28.58
Original2026.06		27.85	78.6	16.23	66.2	0.8506	1.59	29.18
SPARKE2026.06		27.87	77.74	22.2	30.13	0.9739	1.67	29.35
CADS2026.06		39.97	59.14	51.72	59.13	0.5045	2.69	26.89
CADS2026.06		68.13	39.52	57.74	45.25	0.3062	2.32	26.18