AttributeError: 'unet' object has no attribute 'Conv'

[HILALOZTEMEL]

> I think the error message is caused by the yaml file. My input image size is 128 x 128 .

> this is error mesage :

(venv_gandlf) C:\Users\pc\GaNDLF>python gandlf_run -c ./images_and_labels/model.yaml -i ./images_and_labels/train.csv -m ./images_and_labels/model -t True -d cpu WARNING: Initializing 'norm_type' as 'batch' WARNING: Initializing 'verbose' as False WARNING: Initializing 'medcam_enabled' as False WARNING: Initializing 'save_training' as False WARNING: Initializing 'save_output' as False WARNING: Initializing 'in_memory' as False WARNING: Initializing 'pin_memory_dataloader' as False WARNING: Initializing 'enable_padding' as False WARNING: Initializing 'scaling_factor' as 1 WARNING: Initializing 'clip_grad' as None WARNING: Initializing 'track_memory_usage' as False WARNING: Initializing 'print_rgb_label_warning' as True WARNING: Initializing 'data_postprocessing' as {} WARNING: Initializing 'grid_aggregator_overlap' as crop WARNING: Initializing 'clip_mode' as None WARNING: Setting default step_size to: 0.02 Trying to resume training without changing any parameters from previous run. Using previously saved parameter file ./images_and_labels/model\parameters.pkl Using previously saved testing data ./images_and_labels/model\testing_0\data_testing.pkl Using previously saved training+validation data ./images_and_labels/model\testing_0\data_trainAndVal.pkl Number of channels : 1 Constructing queue for train data: 100%|████████████████████████| 643/643 [00:02<00:00, 216.35it/s] C:\Users\pc\anaconda3\envs\venv_gandlf\lib\site-packages\torch\utils\data\dataloader.py:477: UserWarning: This DataLoader will create 16 worker processes in total. Our suggested max number of worker in current system is 4 (cpuset is not taken into account), which is smaller than what this DataLoader is going to create. Please be aware that excessive worker creation might get DataLoader running slow or even freeze, lower the worker number to avoid potential slowness/freeze if necessary. cpuset_checked)) Calculating weights Constructing queue for penalty data: 100%|██████████████████████| 643/643 [00:02<00:00, 241.46it/s] Looping over training data for penalty calculation: 100%|███████| 643/643 [00:05<00:00, 121.90it/s] Constructing queue for validation data: 100%|███████████████████| 161/161 [00:01<00:00, 124.04it/s] All Keys : ['subject_id', '1', 'spacing', 'label', 'path_to_metadata'] Traceback (most recent call last): File "gandlf_run", line 116, in <module> args.reset, File "C:\Users\pc\GaNDLF\GANDLF\cli\main_run.py", line 91, in main_run reset=reset, File "C:\Users\pc\GaNDLF\GANDLF\training_manager.py", line 254, in TrainingManager testing_data=testingData, File "C:\Users\pc\GaNDLF\GANDLF\compute\training_loop.py", line 239, in training_loop ) = create_pytorch_objects(params, training_data, validation_data, device) File "C:\Users\pc\GaNDLF\GANDLF\compute\generic.py", line 69, in create_pytorch_objects model = get_model(parameters) File "C:\Users\pc\GaNDLF\GANDLF\models\__init__.py", line 106, in get_model return global_models_dict[params["model"]["architecture"]](parameters=params) File "C:\Users\pc\GaNDLF\GANDLF\models\unet.py", line 41, in __init__ conv=self.Conv, File "C:\Users\pc\anaconda3\envs\venv_gandlf\lib\site-packages\torch\nn\modules\module.py", line 948, in __getattr__ type(self).__name__, name)) AttributeError: 'unet' object has no attribute 'Conv'

> And this is my yaml file

model.pdf

[sarthakpati]

Your issue is here:

GaNDLF currently only supports 2D and 3D computations.

[sarthakpati]

Not sure what else I can do, since I have provided the model on which your data started training for me. I would suggest take a look at the model config I sent and ensure all previous results have been removed.

[HILALOZTEMEL]

i did the training but output_validation folder is empty ,ı can't solve the reason . What could be the reason?

[sarthakpati]

Unsure what you mean when you say "output_validation is empty". Are you trying to save the output validation masks for every epoch?

[HILALOZTEMEL]

yes , how can i do this?

[sarthakpati]

I am not sure. What do the training/validation logs show?