Thank you for implementing and sharing. I tried to reproduce the training from scratch on the MPI-Sintel-complete dataset and the loss function somehow was not going down. Any advice on how to get it converging would be appreciated.
python train.py --dataset SintelClean --dataset_dir MPI-Sintel-complete
Thank you for implementing and sharing. I tried to reproduce the training from scratch on the MPI-Sintel-complete dataset and the loss function somehow was not going down. Any advice on how to get it converging would be appreciated.
python train.py --dataset SintelClean --dataset_dir MPI-Sintel-complete