ZikangZhou
commented
1 year ago No, I didn't change the hyperparameters except that I use different GPUs for training. As you have changed the batch size, I feel that such a degree of deviation in performance is acceptable.
@ZikangZhou Hi, I followed the script provided in your readme file to reproduce the results, but encountered two issues. Firstly, when I set the batch size to 4 on my 3090 GPU, I encountered out of memory problem. So I adjusted the batch size to 2, while keeping the other hyperparameters unchanged. The resulting reproduction shows that minade=0.732 and minfde=1.294, which has a small deviation from the results reported in the paper. However, there is a significant difference from the pre-trained model that you uploaded, where minade=1.25. I would like to ask if the hyperparameters or other training configurations of the pre-trained model in this version have been changed? Thank you very much.