[GH-ISSUE #127] Ragarding training datasets #112

Closed
opened 2026-10-31 21:11:18 +00:00 by bsakaguchi · 1 comment
Owner

Originally created by @juhongm999 on GitHub (Jul 13, 2021).
Original GitHub issue: https://github.com/openai/CLIP/issues/127

First of all, thank you for great work.
Are you planning to provide the web-crawled datasets used for training CLIP?

Originally created by @juhongm999 on GitHub (Jul 13, 2021). Original GitHub issue: https://github.com/openai/CLIP/issues/127 First of all, thank you for great work. Are you planning to provide the web-crawled datasets used for training CLIP?
Author
Owner

@jongwook commented on GitHub (Jul 18, 2021):

Unfortunately, we don't have a plan to do so, in part because we don't have right to redistribute those data. As an alternative, we provide information on the YFCC100M subset data which we have shown in the paper that it can achieve comparable performance with the ResNet-50 encoder.

<!-- gh-comment-id:882028868 --> @jongwook commented on GitHub (Jul 18, 2021): Unfortunately, we don't have a plan to do so, in part because we don't have right to redistribute those data. As an alternative, we provide information on [the YFCC100M subset data](https://github.com/openai/CLIP/blob/main/data/yfcc100m.md) which we have shown in the paper that it can achieve comparable performance with the ResNet-50 encoder.
Sign in to join this conversation.
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
openai/CLIP#112
No description provided.