How to use multiple GPUs for training？

[heyknow]

[jisnoo123]

Import torch.nn

Wrap the model in nn.DataParallel() like this below, classifier = torch.nn.DataParallel(model.get_model(num_class, normal_channel=args.use_normals))

In terminal, export CUDA_VISIBLE_DEVICES=<GPU numbers you want to use seperated by commas>

Now run train_classification.py