Sample stimuli

How to use

from brainscore_vision import load_benchmark
benchmark = load_benchmark("Maniquet2024-confusion_similarity")
score = benchmark(my_model)

Benchmark API

Code examples

Model scores

Score Legend Min Alignment Max Alignment

Rank	Model	Score
1	resnet_50_v1	.928
2	mobilenet_v2_0_75_160	.928
3	shufflenet_v2_x1_0	.917
4	mobilenet_v2_1_0_128	.892
5	ReAlnet07_cornet	.873
6	ReAlnet07	.873
7	resnet_SIN_IN_FT_IN	.858
8	deit_small_imagenet_100_seed-0	.850
9	ReAlnet09_cornet	.832
10	ReAlnet09	.832
11	ReAlnet10	.832
12	ReAlnet10_cornet	.832
13	resnet_101_v1	.831
14	vit_large_patch14_clip_224:openai_ft_in1k	.826
15	vonegrcnn_62e_nobn	.823
16	yudixie_resnet18_random_0_240719	.820
17	ReAlnet08_cornet	.813
18	ReAlnet08	.813
19	ReAlnet_incorrect	.808
20	ReAlnet01_cornet	.808
21	ReAlnet01	.808
22	VOneCORnet-S	.804
23	vonegrcnn_47e	.803
24	ReAlnet03	.798
25	ReAlnet03_cornet	.798
26	densenet_201_pytorch	.794
27	vonegrcnn_52e_full	.793
28	resnet50-SIN_IN_IN	.785
29	resnet-18_test_m	.763
30	ReAlnet06	.759
31	ReAlnet06_cornet	.759
32	antialias-resnet152	.753
33	ReAlnet02_cornet	.753
34	ReAlnet02	.753
35	ReAlnet05_cornet	.751
36	ReAlnet05	.751
37	yudixie_resnet18_imagenet1kpret_0_240719	.748
38	yudixie_resnet50_imagenet1kpret_0_240908	.743
39	ReAlnet04_cornet	.738
40	ReAlnet04	.738
41	mobilenet_v2_1_0_160	.737
42	resnet50_tutorial	.736
43	vgg_19	.729
44	grcnn	.708
45	effnetb1_272x240	.708
46	resnet50_imagenet_100_seed-0	.679
47	mobilenet_v2_1_0_224	.677
48	deit_small_imagenet_full_seed-0	.669
49	convnext_xlarge:fb_in22k_ft_in1k	.665
50	antialiased-r50	.662
51	resnet50_robust_l2_eps1	.651
52	vgg_16	.645
53	convnext_tiny:in12k_ft_in1k	.640
54	mobilenet_v2_1_4_224	.632
55	mobilenet_v2_1-4_224_pytorch	.632
56	Res2Net50_26w_4s	.630
57	artResNet18_1	.626
58	yudixie_resnet50_random_0_240908	.603
59	vit_relpos_base_patch32_plus_rpn_256:sw_in1k	.600
60	yudixie_resnet18_category_class_0_240719	.598
61	yudixie_resnet50_object_class_0_240908	.586
62	alexnet_training_seed_03	.571
63	yudixie_resnet50_distance_translation_rotation_0_240908	.562
64	cvt_cvt-21-384-in22k_finetuned-in1k_4_LucyV4	.562
65	cvt_cvt-13-384-in22k_finetuned-in1k_4_LucyV4	.559
66	convnext_small_imagenet_10_seed-0	.555
67	alexnet_training_seed_10	.555
68	convnext_large:fb_in22k_ft_in1k	.553
69	vit_tiny_r_s16_p8_384:augreg_in21k_ft_in1k	.553
70	resnet18_ecoset_full	.547
71	inception_v4	.546
72	vit_base_patch16_clip_224:openai_ft_in1k	.540
73	resnet18_imagenet21kP	.538
74	voneresnet-50	.535
75	regnet_y_400mf	.533
76	deit_small_imagenet_1_seed-0	.533
77	alexnet_training_seed_01	.530
78	voneresnet-50-non_stochastic	.529
79	vit_base_patch16_clip_224:openai_ft_in12k_in1k	.524
80	resnet50-VITO-8deg-cc	.520
81	resnet50-vitoimagevidnet8	.520
82	alexnet_training_seed_07	.516
83	nasnet_large	.513
84	efficientnet_b2_imagenet_full	.508
85	yudixie_resnet18_translation_rotation_0_240719	.508
86	alexnet_training_seed_08	.502
87	resnet_152_v1	.500
88	effnetb1_cutmix_augmix_sam_e1_5avg_424x377	.498
89	yudixie_resnet50_category_class_0_240908	.498
90	swin_small_patch4_window7_224:ms_in22k_ft_in1k	.496
91	deit_base_imagenet_full_seed-0	.496
92	resnet152_imagenet_full	.495
93	resnet50_imagenet_10_seed-0	.490
94	alexnet_training_seed_09	.488
95	yudixie_resnet18_cat_obj_class_all_latents_0_240719	.487
96	yudixie_resnet50_distance_rotation_0_240908	.486
97	convnext_femto_ols:d1_in1k	.480
98	resnet50-SIN_IN	.473
99	alexnet_training_seed_06	.473
100	vit_relpos_base_patch16_clsgap_224:sw_in1k	.472
101	alexnet_ks_torevert	.460
102	efficientnet_b0_imagenet_full	.453
103	alexnet_training_seed_04	.450
104	alexnet_training_seed_05	.444
105	nasnet_mobile	.438
106	densenet-201	.436
107	effnetb1_cutmixpatch_SAM_robust32_avge6e8e9e10_manylayers_324x288	.436
108	convnext_small_imagenet_100_seed-0	.435
109	deit_large_imagenet_full_seed-0	.428
110	cvt_cvt-21-384-in1k_4_LucyV4	.424
111	yudixie_resnet18_rotation_reg_0_240719	.419
112	yudixie_resnet18_object_class_0_240719	.418
113	alexnet_training_seed_02	.418
114	vit_large_patch14_clip_224:laion2b_ft_in1k	.416
115	effnetb1_cutmixpatch_augmix_robust32_avge4e7_manylayers_324x288	.412
116	resnet_152_v2	.410
117	resnet34_ecoset_full	.407
118	yudixie_resnet50_cat_obj_class_all_latents_0_240908	.407
119	convnext_small_imagenet_full_seed-0	.395
120	xception	.392
121	pnasnet_large_pytorch	.381
122	convnext_large_imagenet_full_seed-0	.375
123	resnet50-SIN	.371
124	deit_small_imagenet_10_seed-0	.365
125	yudixie_resnet50_distance_translation_0_240908	.365
126	resnet101_imagenet_full	.362
127	resnet50_imagenet_1_seed-0	.358
128	inception_v1	.351
129	AT_efficientnet-b2	.348
130	AdvProp_efficientnet-b2	.348
131	resnext101_32x16d_wsl	.346
132	cvt_cvt-w24-384-in22k_finetuned-in1k_4	.345
133	resnet50_robust_l2_eps3	.341
134	imagenet_l2_3_0	.341
135	yudixie_resnet50_translation_rotation_0_240908	.341
136	cvt_cvt-13-224-in1k_4_LucyV4	.340
137	yudixie_resnet18_distance_rotation_0_240719	.337
138	yudixie_resnet18_translation_reg_0_240719	.326
139	tv_efficientnet-b1	.326
140	yudixie_resnet50_translation_reg_0_240908	.324
141	inception_v3	.324
142	yudixie_resnet18_distance_translation_rotation_0_240719	.323
143	AlexNet_SIN	.322
144	resnet152_ecoset_full	.314
145	resnet-50-robust	.305
146	yudixie_resnet18_distance_translation_0_240719	.302
147	resnet101_ecoset_full	.298
148	efficientnet_b1_imagenet_full	.289
149	resnet50_ecoset_full	.287
150	blt_vs	.284
151	yudixie_resnet50_distance_reg_0_240908	.280
152	efficientnet-b7	.277
153	BiT-S-R101x1	.270
154	inception_v3_pytorch	.262
155	cvt_cvt-13-384-in1k_4_LucyV4	.255
156	AdvProp_efficientnet-b4	.247
157	resnet_101_v2	.239
158	voneresnet-50-robust	.238
159	yudixie_resnet50_rotation_reg_0_240908	.232
160	grcnn_109	.227
161	resnext101_32x32d_wsl	.220
162	resnet-152_v2_pytorch	.212
163	AdvProp_efficientnet-b6	.206
164	efficientnet_b0	.197
165	BiT-S-R50x1	.195
166	resnext101_32x8d_wsl	.195
167	convnext_tiny_imagenet_full_seed-0	.194
168	resnet_50_v2	.189
169	convnext_small_imagenet_1_seed-0	.186
170	yudixie_resnet18_distance_reg_0_240719	.167
171	resnet50_eMMCR_Vanilla	.164
172	convnext_base_imagenet_full_seed-0	.162
173	vit_large_patch14_clip_224:openai_ft_in12k_in1k	.156
174	vit_large_patch14_clip_224:laion2b_ft_in12k_in1k	.138
175	resnet50-vicregl0p9	.123
176	resnext101_32x48d_wsl	.106
177	cvt_cvt-21-224-in1k_4_LucyV4	.100
178	vit_huge_patch14_clip_224:laion2b_ft_in12k_in1k	.098
179	focalnet_tiny_lrf_in1k	.092
180	convnext_tiny_sup	.062
181	resnet50-vicreg	.019
182	resnet50-vicregl0p75	-0.002
183	resnet50-moclr8deg	-0.009
184	resnet50-barlow	-0.065
185	convnext_xxlarge:clip_laion2b_soup_ft_in1k	X
186	MIM	X
187	AdvProp_efficientnet-b8	X
188	BiT-S-R152x4	X
189	vit_large_patch14_clip_336:laion2b_ft_in1k	X
190	BiT-S-R101x3	X
191	BiT-S-R152x2	X
192	TAU	X
193	mobilenet_v2_0_75_224	X
194	SimVP	X
195	omnivore_swinB	X
196	custom_model_cv_18_dagger_408	X
197	PredRNN	X
198	hmax	X
199	pnasnet_large	X
200	densenet-169	X
201	ConvLSTM	X
202	densenet-121	X
203	vit_huge_patch14_clip_336:laion2b_ft_in12k_in1k	X
204	omnivore_swinS	X
205	ViT_L_32_imagenet1k	X
206	convnext_base:clip_laiona_augreg_ft_in1k_384	X
207	antialiased-rnext101_32x8d	X
208	mobilenet_v2_0_5_192	X
209	mobilenet_v2_0_5_224	X
210	mobilenet_v2_0_75_192	X
211	AdvProp_efficientnet-b7	X
212	resnet50-sup	X
213	mobilenet_v2_1_0_192	X
214	vit_large_patch14_clip_336:openai_ft_in12k_in1k	X
215	convnext_large_mlp:clip_laion2b_augreg_ft_in1k_384	X
216	BiT-S-R50x3	X
217	mobilenet_v2_1_3_224	X

Benchmark bibtex

@article {Maniquet2024.04.02.587669,
	author = {Maniquet, Tim and de Beeck, Hans Op and Costantino, Andrea Ivan},
	title = {Recurrent issues with deep neural network models of visual recognition},
	elocation-id = {2024.04.02.587669},
	year = {2024},
	doi = {10.1101/2024.04.02.587669},
	publisher = {Cold Spring Harbor Laboratory},
	URL = {https://www.biorxiv.org/content/early/2024/04/10/2024.04.02.587669},
	eprint = {https://www.biorxiv.org/content/early/2024/04/10/2024.04.02.587669.full.pdf},
	journal = {bioRxiv}
}

Ceiling

Not available

Data: Maniquet2024

Metric: confusion_similarity