Really nice work! I am considering conducting a follow-up to your excellent work but limited by my computation resources. Do you have an estimate of the training time with 8xA100 GPU and how many epochs would be sufficient for an ablation study (to get useful signals)? Thank you!
Really nice work! I am considering conducting a follow-up to your excellent work but limited by my computation resources. Do you have an estimate of the training time with 8xA100 GPU and how many epochs would be sufficient for an ablation study (to get useful signals)? Thank you!