Closed nitba closed 3 years ago
Hi @abelfz
I have the checkpoints for the "decoupled" paper, but not those for the "MaCow" paper. For the CelebA-HQ dataset, it contains totally 30000 samples and I split to 27000/3000 for training and validation.
@XuezheMax Sorry, I feel a bit confused on the dataset, since I split "CelebAMask-HQ" dataset according to the instructions, I have checked previous SOTA on "Celeb-HQ", they shared dataset in tf-records formats (find it hard to check ids). Would you please share the dataset (or val ids) ?
For the CelebA-HQ, I also used the tf-records to extract the images.
Hi @XuezheMax ,
Thanks for your nice works, for the "MaCow" paper, would you please share checkpoints for "CelebHQ dataset"? to have "Celeb-HQ" images, I have used "CelebAMask-HQ" dataset which has 2993 validation and 2824 test samples. is that correct?