|
--- |
|
license: mit |
|
task_categories: |
|
- zero-shot-classification |
|
- image-to-text |
|
language: |
|
- en |
|
pretty_name: yfcc15m |
|
size_categories: |
|
- 10M<n<100M |
|
--- |
|
|
|
## YFCC15M subset used for VLMs |
|
|
|
This dataset contains the ~15M subset of YFCC100M used for training the models in the paper [Quality Not Quantity: On the Interaction between Dataset Design and Robustness of CLIP](https://arxiv.org/abs/2208.05516). The metadata provided in this repo contains both the page-urls and image-download-urls for downloading the dataset. |
|
This dataset can be easily downloaded with [img2dataset](https://github.com/rom1504/img2dataset): |
|
```bash |
|
img2dataset --url_list yfcc15m_final_split_pageandimageurls.csv --input_format "csv" --output_format webdataset --output_folder images --processes_count 2 --thread_count 8 --resize_mode no --enable_wandb True |
|
``` |
