@muellerzr The actual dataset has over 1 million for training and around 130k for validation. You can use a smaller dls instead.
I’ll also remove the wandb callback and let you know.
1 Like
@muellerzr The actual dataset has over 1 million for training and around 130k for validation. You can use a smaller dls instead.
I’ll also remove the wandb callback and let you know.