Hi, As training is quite long, especially with the size of the training set compared to last year, I would like to use multiple GPUs. When ensuring that they are visible with ``` os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID" os.environ["CUDA_VISIBLE_DEVICES"]="0,1,2,3" device = "cuda" ``` I get the following error: ``` Traceback (most recent call last): File "./FeTS_Challenge.py", line 605, in restore_from_checkpoint_folder = restore_from_checkpoint_folder) File "/gpfsdswork/projects/rech/geh/uzq69ur/Challenge/Task_1/fets_challenge/experiment.py", line 290, in run_challenge_experiment task_runner = copy(plan).get_task_runner(collaborator_data_loaders[col]) File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/openfl/federated/plan/plan.py", line 389, in get_task_runner self.runner_ = Plan.build(**defaults) File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/openfl/federated/plan/plan.py", line 182, in build instance = getattr(module, class_name)(**settings) File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/openfl/federated/task/runner_fets_challenge.py", line 43, in __init__ model, optimizer, train_loader, val_loader, scheduler, params = create_pytorch_objects(fets_config_dict, train_csv=train_csv, val_csv=val_csv, device=device) File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/GANDLF/compute/generic.py", line 78, in create_pytorch_objects model, amp=parameters["model"]["amp"], device=device, optimizer=optimizer File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/GANDLF/utils/tensor.py", line 159, in send_model_to_device model = nn.DataParallel(model, "[" + dev + "]") File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 138, in __init__ self.device_ids = [_get_device_index(x, True) for x in device_ids] File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 138, in self.device_ids = [_get_device_index(x, True) for x in device_ids] File "/linkhome/rech/gencre01/uzq69ur/.conda/envs/fets_2022/lib/python3.7/site-packages/torch/_utils.py", line 479, in _get_device_index device = torch.device(device) RuntimeError: Invalid device string: '[' ``` Is it possible to use multiple GPUs with the proposed code for Task 1 in this Challenge? Am I doing something wrong? Thank you, Matthis.