Share your thoughts, 1 month free Claude Pro on usSee more

Class-conditional Image Generation on ImageNet (val)

2.226FID

DDIM-50 steps

Updated 24d ago

Evaluation Results

Method
DDIM-50 steps 2026.02	2.226	241.1	-	-	-	-	4.275	-	-	-	-
Origin 2025.12	2.23	275.64	-	-	83	58	4.57	-	1	0	-
Origin 2025.12	2.25	239.93	-	-	80	59	4.33	-	1	0	-
StyleGAN-XL 2022.06	2.3	262.1	-	-	78	53	-	-	-	-	-
Uniform 50% steps 2025.12	2.42	270.4	-	-	82	57	5.04	8.55	0.14	0.78	-
DPCache 2026.02	2.562	230.6	-	-	-	-	4.503	-	-	-	-
LazyDiT 2025.12	2.7	237.03	-	-	80	59	4.47	-	-	-	-
FORA 2025.12	2.8	-	-	-	80	59	6.21	-	-	-	-
ToCa(N6) 2025.12	3.08	246.59	-	-	79	59	6.58	20.92	0.71	0.21	-
TaylorSeer(N6) + CEM 2025.12	3.08	231.1	-	-	80	57	6.43	25.64	0.83	0.1	-
ToCa(N6) + CEM 2025.12	3.09	248.58	-	-	80	59	6	22.63	0.76	0.16	-
SpeCa 2026.02	3.185	224.24	-	-	-	-	5.69	-	-	-	-
DPCache 2026.02	3.285	215.99	-	-	-	-	5.063	-	-	-	-
Contextual RQ-Transformer 2022.06	3.41	224.6	-	-	79	54	-	-	-	-	-
Contextual RQ-Transformer 2022.06	3.45	221.9	-	-	82	52	-	-	-	-	-
40% steps 2026.02	3.504	221.45	-	-	-	-	4.918	-	-	-	-
TaylorSeer(N6) 2025.12	3.56	223.83	-	-	79	56	7.52	24.69	0.8	0.13	-
RQ-Transformer 2022.06	3.8	323.7	-	-	82	50	-	-	-	-	-
LlamaGen-XL + Reward-based training 2026.03	3.82	195.58	-	-	77	67	-	-	-	-	23.7
LlamaGen-L + Reward-based training 2026.03	3.83	215.9	-	-	79	63	-	-	-	-	23.6
DuCa(N5) + CEM 2025.12	3.96	218.66	-	-	78	55	5.87	23	0.76	0.16	-
LlamaGen-XL 2026.03	3.96	187.75	-	-	74	67	-	-	-	-	23.46
Uniform 33% steps 2025.12	4.24	214.35	-	-	77	56	5.52	9.22	0.17	0.81	-
ADM-G 2022.06	4.59	186.7	-	-	82	52	-	-	-	-	-
AdaCache 2025.12	4.64	-	-	-	-	-	-	-	-	-	-
LlamaGen-L 2026.03	4.64	196.78	-	-	78	63	-	-	-	-	23.5
ToCa(N5) + CEM 2025.12	4.68	212.13	-	-	77	55	6.41	21.59	0.72	0.2	-
SpeCa 2026.02	4.705	203.17	-	-	-	-	7.835	-	-	-	-
TaylorSeer 2026.02	4.776	201.94	-	-	-	-	7.793	-	-	-	-
30% steps 2026.02	5.138	205.65	-	-	-	-	6.12	-	-	-	-
Origin 2025.12	5.31	245.85	-	-	81	68	17.61	-	1	0	-
Q-DiT 2025.12	5.44	237.34	-	-	80	68	17.61	31.1	0.95	0.04	-
Contextual RQ-Transformer 2022.06	5.45	172.6	-	-	81	49	-	-	-	-	-
Q-DiT + CEM 2025.12	5.51	240.36	-	-	80	68	17.49	31.06	0.93	0.05	-
DuCa(N5) 2025.12	6.07	199.64	-	-	74	52	6.64	16.63	0.53	0.39	-
TaylorSeer 2026.02	6.163	186.55	-	-	-	-	9.503	-	-	-	-
MaskGIT 2022.06	6.18	182.1	-	-	80	51	-	-	-	-	-
Q-DiT + CEM 2025.12	6.2	213.5	-	-	76	69	17.62	24.99	0.82	0.14	-
Q-DiT 2025.12	6.31	209.3	-	-	76	69	17.81	24.88	0.82	0.14	-
LlamaGen-B + Reward-based training 2026.03	6.31	163.37	-	-	82	54	-	-	-	-	23.47
ToCa(N5) 2025.12	6.37	199.48	-	-	74	53	7.09	16.56	0.53	0.4	-
IC-GAN 2021.09	6.7	45.9	-	-	-	-	-	-	-	-	-
BigGAN-deep 2022.06	6.95	202.6	-	-	87	23	-	-	-	-	-
LlamaGen-B 2026.03	7.06	119.52	-	-	74	61	-	-	-	-	22.62
IC-GAN 2021.09	7.5	152.6	-	-	-	-	-	-	-	-	-
RQ-Transformer 2022.06	7.55	134	-	-	73	58	-	-	-	-	-
LDM-8-G 2022.06	7.76	209.52	-	-	84	35	-	-	-	-	-
BigGAN 2021.09	8	107.2	-	-	-	-	-	-	-	-	-
BigGAN 2021.09	8	139.1	-	-	-	-	-	-	-	-	-
BigGAN 2021.09	8.1	144.2	-	-	-	-	-	-	-	-	-
IC-GAN 2021.09	8.2	173.8	-	-	-	-	-	-	-	-	-
BigGAN 2021.09	8.3	125	-	-	-	-	-	-	-	-	-
IC-GAN 2021.09	8.3	143.7	-	-	-	-	-	-	-	-	-
IC-GAN 2021.09	8.5	39.7	-	-	-	-	-	-	-	-	-
RQ-Transformer 2022.06	8.71	119	-	-	71	58	-	-	-	-	-
BigGAN 2021.09	9.4	98.7	-	-	-	-	-	-	-	-	-
IC-GAN 2021.09	9.5	108.6	-	-	-	-	-	-	-	-	-
BigGAN 2021.09	10.2	30.1	-	-	-	-	-	-	-	-	-
IC-GAN 2021.09	10.6	100.1	-	-	-	-	-	-	-	-	-
ADM 2022.06	10.94	101	-	-	69	63	-	-	-	-	-
VQ-Diffusion 2022.06	11.89	-	-	-	-	-	-	-	-	-	-
BigGAN 2021.09	12.3	27	-	-	-	-	-	-	-	-	-
LDM-8 2022.06	15.51	79.03	-	-	65	63	-	-	-	-	-
VQ-GAN 2022.06	15.78	74.3	-	-	-	-	-	-	-	-	-
ReACGAN 2021.11	16.32	50.3	23.21	51.602	-	-	-	-	-	-	-
ImageBART 2022.06	21.19	61.6	-	-	-	-	-	-	-	-	-
BigGAN 2021.11	24.68	28.63	29.994	53.842	-	-	-	-	-	-	-
ContraGAN 2021.11	25.16	25.25	2.866	11.482	-	-	-	-	-	-	-
ACGAN 2021.11	26.35	62.99	62.412	84.899	-	-	-	-	-	-	-
Real data (validation) 2021.11	-	173.33	70.822	89.574	-	-	-	-	-	-	-