synset_id
int64 4.48k
15.1M
| image_id
int64 1
174k
|
|---|---|
13,145,040 | 48 |
13,145,040 | 52 |
13,145,040 | 59 |
13,145,040 | 84 |
13,145,040 | 90 |
13,145,040 | 110 |
13,145,040 | 145 |
13,145,040 | 171 |
13,145,040 | 174 |
13,145,040 | 175 |
13,145,040 | 198 |
13,145,040 | 210 |
13,145,040 | 223 |
13,145,040 | 264 |
13,145,040 | 274 |
13,145,040 | 277 |
13,145,040 | 300 |
13,145,040 | 315 |
13,145,040 | 316 |
13,145,040 | 333 |
13,145,040 | 349 |
13,145,040 | 376 |
13,145,040 | 387 |
13,145,040 | 394 |
13,145,040 | 396 |
13,145,040 | 401 |
13,145,040 | 406 |
13,145,040 | 438 |
13,145,040 | 440 |
13,145,040 | 441 |
13,145,040 | 454 |
13,145,040 | 479 |
13,145,040 | 483 |
13,145,040 | 495 |
13,145,040 | 497 |
13,145,040 | 499 |
13,145,040 | 505 |
13,145,040 | 516 |
13,145,040 | 527 |
13,145,040 | 528 |
13,145,040 | 534 |
13,145,040 | 539 |
13,145,040 | 548 |
13,145,040 | 561 |
13,145,040 | 562 |
13,145,040 | 573 |
13,145,040 | 576 |
13,145,040 | 598 |
13,145,040 | 604 |
13,145,040 | 607 |
13,145,040 | 615 |
13,145,040 | 645 |
13,145,040 | 646 |
13,145,040 | 666 |
13,145,040 | 667 |
13,145,040 | 669 |
13,145,040 | 676 |
13,145,040 | 679 |
13,145,040 | 686 |
13,145,040 | 687 |
13,145,040 | 703 |
13,145,040 | 708 |
13,145,040 | 715 |
13,145,040 | 723 |
13,145,040 | 727 |
13,145,040 | 732 |
13,145,040 | 758 |
13,145,040 | 763 |
13,145,040 | 803 |
13,145,040 | 805 |
13,145,040 | 848 |
13,145,040 | 854 |
13,145,040 | 907 |
13,145,040 | 908 |
13,145,040 | 916 |
13,145,040 | 925 |
13,145,040 | 972 |
13,145,040 | 989 |
13,145,040 | 992 |
13,145,040 | 1,001 |
13,145,040 | 1,018 |
13,145,040 | 1,054 |
13,145,040 | 1,055 |
13,145,040 | 1,059 |
13,145,040 | 1,069 |
13,145,040 | 1,070 |
13,145,040 | 1,097 |
13,145,040 | 1,102 |
13,145,040 | 1,112 |
13,145,040 | 1,113 |
13,145,040 | 1,120 |
13,145,040 | 1,126 |
13,145,040 | 1,128 |
13,145,040 | 1,143 |
13,145,040 | 1,149 |
13,145,040 | 1,167 |
13,145,040 | 1,176 |
13,145,040 | 1,183 |
13,145,040 | 1,207 |
13,145,040 | 1,214 |
This dataset contains the embeddings of each image in ImageNet. The embeddings were produced by the OpenAI implementation of CLIP with the ViT-L/14 variant using the snapshot at https://openaipublic.azureedge.net/clip/models/b8cca3fd41ae0c99ba7e8951adf17d267cdb84cd88be6f7c2e0eca1737a03836/ViT-L-14.pt.
The version of ImageNet used was the full set in the Winter 2021 release, MD5: ab313ce03179fd803a401b02c651c0a2.
There are 13,158,856 images in the dataset. The file index.parquet has two columns: synset_id and image_id.
Each image has an entry in index.parquet. Their embeddings are stored in a single file named imagenet_ViT_L_14.npy, which is a (13,158,856, 768)-shaped matrix. The ordering of the entries in index.parquets is exactly the same that in the .npy file.
- Downloads last month
- 35