CLIPSelf
CLIPSelf copied to clipboard
Providing Scripts and Checkpoints for training on CC3M
First of all, thank you very much for publishing code that accompanies your paper.
I was wondering if you would be willing to share the training script as well as a pre-trained model checkpoint for training on GCC3M (Table 8 of the paper)?
Thanks in advance!
Hi, to run CLIPSelf on CC3M, simply prepare a json file under the coco format:
{'images': [{'id': xxx}, {'id': xxx}, {'id': xxx}, }
And make sure to download cc3m images in their original sizes.
I erroneously deleted the cc3m checkpoint. I re-ran the experiment (there were only very small fluctuations on the classification mAccs) and it is now available at link.