Run the program with multi GPU server

[L N Sastry Varanasi]

I would like to train the dataset using the following command as given in the README.md

python seq2point_train.py --appliance_name 'kettle' --datadir './dataset_management/refit/' --save_dir './trained_model' --transfer_model False --crop_dataset 10000 --gpus 4

but it forcefully run on a single GPU
Could you please help me how i can train on multi gpu server