Sample stimuli

How to use

from brainscore_vision import load_benchmark
benchmark = load_benchmark("Geirhos2021highpass-error_consistency")
score = benchmark(my_model)

Benchmark API

Code examples

Model scores

Score Legend Min Alignment Max Alignment

Rank	Model	Score
1	cvt_cvt-w24-384-in22k_finetuned-in1k_4	.875
2	vit_large_patch14_clip_224:openai_ft_in12k_in1k	.858
3	convnext_xlarge:fb_in22k_ft_in1k	.851
4	vit_large_patch14_clip_224:laion2b_ft_in12k_in1k	.827
5	vit_large_patch14_clip_224:laion2b_ft_in1k	.812
6	vit_large_patch14_clip_224:openai_ft_in1k	.807
7	convnext_large:fb_in22k_ft_in1k	.806
8	vit_huge_patch14_clip_224:laion2b_ft_in12k_in1k	.793
9	vit_large_patch14_clip_336:laion2b_ft_in1k	.722
10	fixres_resnext101_32x48d_wsl	.716
11	convnext_base:clip_laiona_augreg_ft_in1k_384	.709
12	vit_huge_patch14_clip_336:laion2b_ft_in12k_in1k	.652
13	vit_large_patch14_clip_336:openai_ft_in12k_in1k	.592
14	CLIP_resnet50_float32	.550
15	CLIP_resnet50	.550
16	vit_base_patch16_clip_224:openai_ft_in1k	.506
17	vit_base_patch16_clip_224:openai_ft_in12k_in1k	.493
18	convnext_tiny:in12k_ft_in1k	.457
19	convnext_tiny_imagenet_full_seed-0	.442
20	CLIP-RN50	.422
21	convnext_large_mlp:clip_laion2b_augreg_ft_in1k_384	.419
22	effnetv2m_custom384	.410
23	convnext_xxlarge:clip_laion2b_soup_ft_in1k	.404
24	SWSL_resnext101_32x8d	.380
25	resnext101_32x48d_wsl	.367
26	cvt_cvt-21-384-in22k_finetuned-in1k_4_LucyV4	.364
27	convnext_large_imagenet_full_seed-0	.355
28	convnext_small_imagenet_full_seed-0	.330
29	cvt_cvt-13-384-in1k_4_LucyV4	.305
30	resnet_152_v2	.303
31	resnext101_32x16d_wsl	.294
32	cv_18_dagger_408_pretrained	.293
33	resnext101_32x32d_wsl	.291
34	convnext_base_imagenet_full_seed-0	.287
35	resnet50-VITO-8deg-cc	.285
36	swin_small_patch4_window7_224:ms_in22k_ft_in1k	.262
37	vision_transformer_vit_large_patch16_224	.248
38	convnext_tiny_sup	.226
39	focalnet_tiny_lrf_in1k	.220
40	resnet-152_v2_pytorch	.202
41	cvt_cvt-21-384-in1k_4_LucyV4	.200
42	cvt_cvt-21-224-in1k_4_LucyV4	.195
43	vit_relpos_base_patch16_clsgap_224:sw_in1k	.189
44	pnasnet_large	.177
45	resnext101_32x8d_wsl	.176
46	convnext_femto_ols:d1_in1k	.175
47	SWSL_resnet50	.154
48	cvt_cvt-13-224-in1k_4_LucyV4	.151
49	pnasnet_large_pytorch	.150
50	nasnet_large	.149
51	inception_v3	.143
52	effnetb1_cutmix_augmix_sam_e1_5avg_424x377	.142
53	CLIP_ViT-B_32	.139
54	inception_v4	.139
55	pnasnet_large-tf	.139
56	inception_v3_pytorch	.139
57	inception_v3-tf	.135
58	densenet-169	.132
59	convnext_small_imagenet_100_seed-0	.131
60	cvt_cvt-13-384-in22k_finetuned-in1k_4_LucyV4	.127
61	resnet_101_v2	.123
62	resnet-152_v2-tf	.120
63	xception	.118
64	densenet-169-keras	.116
65	vit_relpos_base_patch32_plus_rpn_256:sw_in1k	.116
66	ViT_B_16_imagenet1k	.111
67	xception-keras	.110
68	resnet34_imagenet_full	.101
69	resnet_50_v2	.100
70	effnetb1_cutmixpatch_augmix_robust32_avge4e7_manylayers_324x288	.100
71	BiT-S-R50x3	.100
72	alexnet_training_seed_05	.096
73	alexnet_training_seed_01	.096
74	tf_efficientnetv2_s_in21ft1k_robust_linf12255_400x400	.095
75	deit_small_imagenet_10_seed-0	.095
76	deit_small_imagenet_1_seed-0	.095
77	resnet50_imagenet_10_seed-0	.095
78	imagenet_l2_10_0	.094
79	resnet50_finetune_cutmix_e3_robust_linf8255_e0_247x234	.093
80	BiT-S-R152x2	.093
81	resnet50_FractalDB	.092
82	resnet50-vicregl0p9	.091
83	barlow-twins-resnet50	.091
84	alexnet_linf_8_robust	.090
85	alexnet_l2_3_robust	.090
86	resnet50_linf_8_robust	.089
87	inception_v4-tf	.089
88	alexnet_training_seed_07	.088
89	effnetb1_272x240	.088
90	alexnet_robust_correct	.086
91	custom_model_cv_18_dagger_408	.085
92	AdvProp_efficientnet-b4	.083
93	AT_efficientnet-b4	.083
94	mobilenet_v2_0_5_224	.082
95	blt_vs	.081
96	vonealexnet_gaussian_noise_std4_fixed	.081
97	mobilenet_v1_0.5_192	.080
98	resnet50_linf_4_robust	.079
99	densenet-201-keras	.078
100	effnetb1_cutmixpatch_augmix_robust32_e5_324x288	.078
101	resnet50_random_l2_perturb	.077
102	resnet50-meshes-lt-100-original-scratch	.075
103	mobilenet_v1_0.25_128	.075
104	alexnet_training_seed_08	.075
105	alexnet_training_seed_04	.075
106	alexnet_training_seed_10	.075
107	artResNet18_1	.075
108	alexnet_training_seed_02	.075
109	resnet101	.075
110	BiT-S-R101x3	.074
111	convnext_small_imagenet_10_seed-0	.073
112	resnet50_byol	.073
113	konkle_alexnetgn_ipcl_ref12_supervised_ipcl_aug	.073
114	texture_shape_alexnet_trained_on_SIN	.073
115	alexnet_training_seed_03	.072
116	mobilenet_v1_0.75_160	.072
117	vgg-19-keras	.072
118	nasnet_large-tf	.071
119	resnet50-vicreg	.071
120	BiT-S-R152x4	.066
121	voneresnet-50-robust	.066
122	vgg-16-keras	.066
123	mobilenet_v1_0.25_192	.065
124	antialias-resnet152	.064
125	mobilenet_v1_1.0_224	.064
126	AdvProp_efficientnet-b8	.064
127	resnet50-moclr8deg	.063
128	mobilenet_v2_0_5_192	.062
129	effnetb1_cutmixrespatch_SAM_robust32_e2_manylayers_324x288	.061
130	resnet_SIN_IN_FT_IN	.060
131	resnet50-SIN_IN_IN	.060
132	resnet50_robust_l2_eps1	.060
133	resnet_101_v1	.059
134	mobilenet_v1_0.75_128	.059
135	alexnet_reduced_aliasing_early_checkpoint	.059
136	konkle_alexnetgn_ipcl_ref01_primary_model	.059
137	resnet50_l2_3_robust	.056
138	AdvProp_efficientnet-b0	.056
139	effnetb1_cutmixpatch_SAM_robust32_avge6e8e9e10_manylayers_324x288	.056
140	AlexNet_SIN	.055
141	resnet50_robust_l2_eps3	.055
142	imagenet_l2_3_0	.055
143	resnet-50-robust	.055
144	mobilenet_v2_1.0_160-tf	.055
145	mobilenet_v1_0.25_160	.055
146	densenet-201	.054
147	densenet_201_pytorch	.054
148	efficientnet_b0	.054
149	mobilenet_v2_1_0_128	.054
150	vonegrcnn_62e_nobn	.053
151	alexnet_training_seed_06	.053
152	AdvProp_efficientnet-b7	.053
153	mobilenet_v1_0.5_224	.053
154	vonegrcnn_52e_full	.052
155	vonegrcnn_47e	.052
156	squeezenet1_0	.051
157	convnext_small_imagenet_1_seed-0	.051
158	CORnet-S	.051
159	mobilenet_v1_1.0_128	.051
160	deit_small_imagenet_100_seed-0	.050
161	mobilenet_v2_1_0_224	.050
162	effnetb1_cutmixrespatch_SAM_robust32_e10_manylayers_324x288	.050
163	ReAlnet07	.050
164	mobilenet_v2_1.0_224-tf	.050
165	mobilevit_small	.049
166	resnet50-SIN	.049
167	ReAlnet06	.048
168	resnet152_imagenet_full	.048
169	ReAlnet08	.047
170	cornet_s	.046
171	ReAlnet10	.046
172	resnet-18	.046
173	resnet50_imagenet_full	.045
174	alexnet_ks_torevert	.045
175	alexnet	.045
176	alexnet	.045
177	alexnet-baseline	.045
178	alexnet	.045
179	alexnet	.045
180	alexnet	.045
181	alexnet	.045
182	alexnet	.045
183	alexnet	.045
184	alexnet	.045
185	mobilenet_v2_0_75_192	.045
186	resnet50-simclr-vissl	.044
187	resnet50-meshes-lt-100-original-pretrained	.044
188	vonegrcnn_52e_full	.043
189	nasnet_mobile	.043
190	mobilenet_v2_1_3_224	.043
191	resnet101_imagenet_full	.043
192	efficientnet-b7	.043
193	vonegrcnn_47e	.043
194	mobilenet_v1_0.5_160	.042
195	resnet50-SIN_IN	.042
196	regnet_y_400mf	.042
197	resnet50_moco_v2	.041
198	nasnet_mobile-tf	.041
199	Res2Net50_26w_4s	.040
200	resnet50	.040
201	mobilenet_v1_1.0_160	.040
202	mobilenet_v2_0.75_96	.039
203	mobilenet_v1_0.5_128	.038
204	mobilenet_v1_0.25_224	.038
205	squeezenet1_1	.038
206	vit_tiny_r_s16_p8_384:augreg_in21k_ft_in1k	.038
207	alexnet_random_linf8_perturb	.037
208	voneresnet-50	.037
209	vonegrcnn_62e_nobn	.037
210	mobilenet_v2_0.75_224-tf	.036
211	inception_v1	.036
212	deit_large_imagenet_full_seed-0	.036
213	mobilenet_v2_1_0_160	.036
214	mobilenet_v2_1.3_224-tf	.035
215	mobilenet_v2_1_0_192	.035
216	alexnet_early_checkpoint	.035
217	mobilenet_v2_1.0_192-tf	.034
218	texture_shape_resnet50_trained_on_SIN	.033
219	alexnet_training_seed_09	.033
220	mobilenet_v2_0_75_160	.033
221	mobilenet_v2_0.5_96	.033
222	resnet-50_untrained	.033
223	ReAlnet05	.032
224	resnet_152_v1	.031
225	resnet50_imagenet_1_seed-0	.031
226	ReAlnet_incorrect	.031
227	ReAlnet01	.031
228	resnet-50-pytorch	.030
229	resnet50-sup	.030
230	resnet50_tutorial	.030
231	resnet_50_v1	.030
232	resnet-34	.030
233	inception_v1-tf	.030
234	BiT-S-R50x1	.028
235	ReAlnet03	.027
236	ReAlnet02	.026
237	mobilenet_v2_0.35_96	.026
238	ViT_B_32_imagenet1k	.024
239	resnet50_random_linf8_perturb	.024
240	BiT-S-R101x1	.023
241	resnet50-barlow	.023
242	alexnet	.023
243	mobilenet_v2_0.75_160-tf	.023
244	mobilenet_v1_0.75_192	.023
245	mobilenet_v2_0.35_128	.022
246	ReAlnet09	.022
247	resnet-50_v1-tf	.022
248	mobilenet_v2_0_75_224	.022
249	antialiased-rnext101_32x8d	.022
250	inception_v2-tf	.022
251	mobilenet_v2_0.5_160	.021
252	mobilenet_v2_1.0_96	.021
253	mobilenet_v1_1.0_192	.020
254	antialiased-r50	.020
255	mobilenet_v2_0.75_192-tf	.019
256	deit_small_imagenet_full_seed-0	.019
257	mobilenet_v2_1.4_224-tf	.019
258	mobilenet_v2_0.5_224-tf	.018
259	resnet50-vicregl0p75	.016
260	resnet50_imagenet_100_seed-0	.015
261	efficientnet_b0_imagenet_full	.015
262	resnet18_imagenet_full	.015
263	resnet-152_v1-tf	.015
264	densenet-121-keras	.015
265	grcnn_109	.015
266	mobilenet_v2_1-4_224_pytorch	.015
267	mobilenet_v2_1_4_224	.015
268	0.5x_resnet-18_LC_w_sh_1_iter	.014
269	AdvProp_efficientnet-b6	.014
270	mobilenet_v1_0.75_224	.014
271	deit_base_imagenet_full_seed-0	.013
272	resnet50_simclr	.013
273	mobilenet_v2_0.5_192-tf	.013
274	mobilenet_v2_0.35_160	.012
275	mobilenet_v2_1.0_128-tf	.012
276	AdvProp_efficientnet-b2	.010
277	AT_efficientnet-b2	.010
278	mobilenet_v2_0.35_192	.010
279	vgg_16	.010
280	mobilenet_v2_0.35_224	.010
281	mobilenet_v2_0.5_128	.010
282	grcnn	.009
283	resnet-50_v2-tf	.007
284	mobilenet_v2_0.75_128	.005
285	resnet-101_v1-tf	.004
286	ReAlnet04	.004
287	alexnet_random_l2_3_perturb	.004
288	shufflenet_v2_x1_0	.004
289	vgg_19	.003
290	densenet-121	.003
291	resnet50-vitoimagevidnet8	.003
292	efficientnet_b2_imagenet_full	.002
293	ViT_L_32_imagenet1k	.002
294	efficientnet_b1_imagenet_full	.001
295	voneresnet-50-non_stochastic	-0.030
296	resnet-18-LC_conv_init_m	X
297	resnet-18-LC_w_sh_1_iter_m	X
298	resnet18-contrastive_multiview	X
299	0.5x_resnet-18	X
300	resnet-18-LC_w_sh_10_iter_conv_init_m	X
301	resnet-18-LC_w_sh_100_iter_m	X
302	ReAlnet01_cornet	X
303	fulltest_microblockvf_nobottleneck_freeat_nopretrain_eps2_m4	X
304	RN50	X
305	0.5x_resnet-18_LC_w_sh_100_iter	X
306	resnet50_ImageNet	X
307	yudixie_resnet18_translation_reg_0_240719	X
308	resnet50_pretrained_with_retinal_waves	X
309	yudixie_resnet18_distance_reg_0_240719	X
310	yudixie_resnet18_rotation_reg_0_240719	X
311	ReAlnet02_cornet	X
312	resnet-18-LC_w_sh_10_iter_m	X
313	yudixie_resnet18_imagenet1kpret_0_240719	X
314	resnet50-cifar	X
315	ReAlnet03_cornet	X
316	RN50	X
317	resnet18_ecoset_full	X
318	resnet34_ecoset_full	X
319	VOneCORnet-S	X
320	resnet50_ecoset_full	X
321	resnet101_ecoset_full	X
322	resnet18-relative_position	X
323	resnet152_ecoset_full	X
324	resnet18-autoencoder	X
325	yudixie_resnet50_rotation_reg_0_240908	X
326	dorinet_cornet_z	X
327	yudixie_resnet50_distance_translation_0_240908	X
328	yudixie_resnet50_distance_rotation_0_240908	X
329	yudixie_resnet18_distance_translation_0_240719	X
330	ReAlnet04_cornet	X
331	yudixie_resnet18_translation_rotation_0_240719	X
332	resnet18-simclr	X
333	yudixie_resnet18_distance_rotation_0_240719	X
334	yudixie_resnet18_category_class_0_240719	X
335	ReAlnet05_cornet	X
336	yudixie_resnet18_random_0_240719	X
337	resnet-18-LC_w_sh_1_iter_conv_init_m	X
338	yudixie_resnet18_cat_obj_class_all_latents_0_240719	X
339	resnet-18_test_m	X
340	yudixie_resnet18_distance_translation_rotation_0_240719	X
341	yudixie_resnet18_object_class_0_240719	X
342	ViT_L_32	X
343	ReAlnet06_cornet	X
344	yudixie_resnet50_translation_reg_0_240908	X
345	BiT-M-R152x2	X
346	yudixie_resnet50_translation_rotation_0_240908	X
347	ViT_B_32	X
348	ReAlnet07_cornet	X
349	resnet18-depth_prediction	X
350	ViT_B_16	X
351	r3m_resnet34_nocrop	X
352	ReAlnet08_cornet	X
353	BiT-M-R50x1	X
354	omnivore_swinB	X
355	resnet18-contrastive_predictive	X
356	BiT-M-R50x3	X
357	ReAlnet09_cornet	X
358	omnivore_swinT	X
359	resnet-18-LC_untrained	X
360	resnet-18_untrained	X
361	resnet-50x2_untrained	X
362	omnivore_swinS	X
363	dcgan	X
364	v1-pyr-nodown	X
365	ReAlnet10_cornet	X
366	unet_entire	X
367	ViT-B/32	X
368	CORnetZ_CIFAR10_bs32_20_04	X
369	openclip	X
370	pixels	X
371	resnet-18-LC_w_sh_100_iter_conv_init_m	X
372	resnet50	X
373	my-model	X
374	0.5x_resnet-18_LC_w_sh_10_iter	X
375	my-model	X
376	vggface	X
377	yudixie_resnet50_object_class_0_240908	X
378	yudixie_resnet50_cat_obj_class_all_latents_0_240908	X
379	resnet18-instance_recognition	X
380	CORnet-Z	X
381	yudixie_resnet50_imagenet1kpret_0_240908	X
382	r3m_resnet18	X
383	yudixie_resnet50_random_0_240908	X
384	yudixie_resnet50_imagenet1kpret_0_240312	X
385	r3m_resnet50	X
386	resnet18-deepcluster	X
387	resnet18-local_aggregation	X
388	resnet18-colorization	X
389	CORnetZ_CIFAR10	X
390	yudixie_resnet50_category_class_0_240908	X
391	r3m_resnet34	X
392	0.5x_resnet-18_LC	X
393	resnet-18-LC_m	X
394	r3m_resnet50_nocrop	X
395	resnet18_imagenet21kP	X
396	resnet18-supervised	X
397	yudixie_resnet50_distance_translation_rotation_0_240908	X
398	ConvLSTM	X
399	PredRNN	X
400	yudixie_resnet50_distance_reg_0_240908	X
401	SimVP	X
402	resnet-18-LC_1st_conv_m	X
403	TAU	X
404	MIM	X
405	hmax	X
406	FrankRobWobv0	X

Benchmark bibtex

@article{geirhos2021partial,
              title={Partial success in closing the gap between human and machine vision},
              author={Geirhos, Robert and Narayanappa, Kantharaju and Mitzkus, Benjamin and Thieringer, Tizian and Bethge, Matthias and Wichmann, Felix A and Brendel, Wieland},
              journal={Advances in Neural Information Processing Systems},
              volume={34},
              year={2021},
              url={https://openreview.net/forum?id=QkljT4mrfs}
        }

Ceiling

0.44.

Note that scores are relative to this ceiling.

Data: Geirhos2021highpass

Metric: error_consistency