Skip to content

Instantly share code, notes, and snippets.

@rwightman
Last active May 25, 2023 23:07
Show Gist options
  • Save rwightman/26043466fa319a7e3b920ae5faae97ef to your computer and use it in GitHub Desktop.
Save rwightman/26043466fa319a7e3b920ae5faae97ef to your computer and use it in GitHub Desktop.
model image_size embed_dim gmacs macts mparams image_gmacs image_macts image_mparams text_gmacs text_macts text_mparams
ViT-S-32-alt 224 256 1.78 4.71 43.22 1.15 2.5 22.59 0.64 2.21 20.63
ViT-S-32 224 384 2.84 6.48 63.09 1.15 2.5 22.64 1.69 3.98 40.44
ViT-M-32-alt 224 384 3.69 7.31 80.07 2.0 3.34 39.63 1.69 3.98 40.44
ViT-M-32 224 512 4.98 8.64 103.12 2.0 3.34 39.69 2.98 5.3 63.43
ViT-S-16-alt 224 256 5.25 14.16 42.4 4.61 11.95 21.76 0.64 2.21 20.63
ViT-S-16 224 384 6.3 15.92 62.26 4.61 11.95 21.81 1.69 3.98 40.44
ViT-B-32-quickgelu 224 512 7.4 10.31 151.28 4.41 5.01 87.85 2.98 5.3 63.43
ViT-B-32 224 512 7.4 10.31 151.28 4.41 5.01 87.85 2.98 5.3 63.43
convnext_tiny 224 1024 7.46 18.74 92.3 4.47 13.44 28.61 2.98 5.3 63.69
RN50 224 1024 9.12 18.29 102.01 6.14 12.98 38.32 2.98 5.3 63.69
RN50-quickgelu 224 1024 9.12 18.29 102.01 6.14 12.98 38.32 2.98 5.3 63.69
ViT-M-16-alt 224 384 9.7 19.91 78.98 8.0 15.93 38.53 1.69 3.98 40.44
ViT-M-16 224 512 10.99 21.23 102.02 8.0 15.93 38.59 2.98 5.3 63.43
vit_relpos_medium_patch16_cls_224 224 512 11.01 23.54 101.94 8.03 18.24 38.51 2.98 5.3 63.43
mt5-base-ViT-B-32 224 512 11.06 13.9 365.71 4.41 5.01 87.85 6.65 8.9 277.86
roberta-ViT-B-32 224 512 11.07 12.96 212.72 4.41 5.01 87.85 6.66 7.95 124.87
xlm-roberta-base-ViT-B-32 224 512 11.07 12.96 366.12 4.41 5.01 87.85 6.66 7.95 278.27
convnext_small 224 512 11.69 26.86 113.28 8.71 21.56 49.85 2.98 5.3 63.43
ViT-B-32-plus-256 256 640 12.43 14.38 210.3 7.79 7.76 119.13 4.64 6.63 91.16
RN101 224 512 12.79 23.38 119.69 9.81 18.08 56.26 2.98 5.3 63.43
RN101-quickgelu 224 512 12.79 23.38 119.69 9.81 18.08 56.26 2.98 5.3 63.43
vit_medium_patch16_gap_256 256 512 13.57 27.45 102.04 10.59 22.15 38.61 2.98 5.3 63.43
convnext_base 224 512 18.37 34.05 151.52 15.38 28.75 88.09 2.98 5.3 63.43
swin_base_patch4_window7_224 224 640 20.11 43.25 178.56 15.47 36.63 87.4 4.64 6.63 91.16
ViT-B-16 224 512 20.57 29.2 149.62 17.58 23.9 86.19 2.98 5.3 63.43
convnext_base_w 256 640 24.73 44.18 179.39 20.09 37.55 88.22 4.64 6.63 91.16
RN50x4 288 640 25.99 41.9 178.3 21.35 35.27 87.14 4.64 6.63 91.16
ViT-B-16-plus 224 640 28.41 34.5 208.35 23.77 27.88 117.19 4.64 6.63 91.16
ViT-B-16-plus-240 240 640 32.05 39.71 208.38 27.41 33.08 117.21 4.64 6.63 91.16
convnext_base_w_320 320 640 36.03 65.3 179.39 31.39 58.68 88.22 4.64 6.63 91.16
convnext_large 224 768 41.06 51.08 321.06 34.4 43.13 197.41 6.66 7.95 123.65
convnext_large_d 256 768 53.81 66.93 351.77 44.94 56.33 199.77 8.88 10.6 152.0
ViT-L-16 224 768 68.26 71.47 427.74 61.6 63.52 304.09 6.66 7.95 123.65
convnext_large_d_320 320 768 79.09 98.62 351.77 70.21 88.02 199.77 8.88 10.6 152.0
RN50x16 384 768 81.55 111.49 290.98 74.9 103.54 167.33 6.66 7.95 123.65
ViT-L-14 224 768 87.73 96.74 427.62 81.08 88.79 303.97 6.66 7.95 123.65
convnext_xlarge 256 1024 95.36 89.24 603.51 79.65 75.11 350.25 15.71 14.13 253.26
convnext_large_d_384 384 768 109.98 137.34 351.77 101.11 126.74 199.77 8.88 10.6 152.0
ViT-L-14-280 280 768 136.0 168.66 427.76 129.34 160.71 304.11 6.66 7.95 123.65
ViT-L-16-320 320 768 136.07 168.66 427.95 129.42 160.71 304.3 6.66 7.95 123.65
ViT-H-16 224 1024 150.96 122.01 986.26 127.4 100.81 632.23 23.57 21.2 354.03
xlm-roberta-large-ViT-H-14 224 1024 190.97 160.62 1193.01 167.4 139.41 632.08 23.57 21.2 560.94
ViT-H-14 224 1024 190.97 160.61 986.11 167.4 139.41 632.08 23.57 21.2 354.03
ViT-L-14-336 336 768 197.76 278.19 427.94 191.1 270.24 304.29 6.66 7.95 123.65
convnext_xxlarge 256 1024 221.66 145.66 1200.58 198.09 124.45 846.54 23.57 21.2 354.03
mt5-xl-ViT-H-14 224 1024 257.13 185.6 2306.75 167.4 139.41 632.08 89.72 46.19 1674.68
RN50x64 448 1024 276.8 249.73 623.26 265.02 239.13 420.38 11.78 10.6 202.88
ViT-g-14 224 1024 290.74 213.84 1366.68 267.18 192.64 1012.65 23.57 21.2 354.03
convnext_xxlarge_320 320 1024 333.08 215.66 1200.58 309.52 194.46 846.54 23.57 21.2 354.03
ViT-H-14-336 336 1024 414.53 428.74 986.52 390.97 407.54 632.49 23.57 21.2 354.03
ViT-bigG-14 224 1280 532.92 310.71 2539.57 483.96 275.37 1844.91 48.96 35.34 694.66
ViT-G-14 224 1280 532.92 310.71 2539.57 483.96 275.37 1844.91 48.96 35.34 694.66
ViT-g-14-336 336 1024 644.21 571.87 1367.13 620.65 550.67 1013.1 23.57 21.2 354.03
ViT-H-8-256 256 1024 755.19 980.27 986.59 731.62 959.07 632.55 23.57 21.2 354.03
ViT-e-14 224 1280 1046.02 475.2 4581.09 990.94 435.44 3807.72 55.08 39.75 773.37
ViT-G-14-336 336 1280 1165.0 795.39 2540.1 1116.04 760.05 1845.44 48.96 35.34 694.66
ViT-e-14-336 336 1280 2316.93 1182.82 4581.66 2261.85 1143.07 3808.29 55.08 39.75 773.37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment