Datasets:
Train issue: AssertionError
Cell In[5], line 3
1 # reduce training epochs for demo
2 args['epochs'] = 25
----> 3 training_script.main(**vars(args))
File c:\Users\leduc\Downloads\phileo-bench-main\training_script.py:470, in main(experiment_name, downstream_task, model_name, augmentations, batch_size, model_device, generator_device, num_workers, early_stop, epochs, input_channels, output_channels, input_size, lr, lr_scheduler, n_shot, split_ratio, regions, vis_val, warmup, warmp_steps, warmup_gamma, pretrained_model_path, freeze_pretrained, data_path_128_10m, data_path_224_10m, data_path_224_30m, output_path, data_parallel, device_ids)
467 if isinstance(n_shot, int):
468 OUTPUT_FOLDER = f'{OUTPUT_FOLDER}{n_shot}'
--> 470 x_train, y_train, x_val, y_val = data_protocol.protocol_fewshot_memmapped(
471 folder=dataset_folder,
472 dst='/mnt/d/downstream_datasets_nshot/',
473 n=n_shot,
474 regions=regions,
475 y=downstream_task,
476 data_selection='create',
477 name=dataset_name)
479 elif isinstance(split_ratio, float):
480 OUTPUT_FOLDER = f'{OUTPUT_FOLDER}{split_ratio}'
File c:\Users\leduc\Downloads\phileo-bench-main\utils\data_protocol.py:468, in protocol_fewshot_memmapped(folder, dst, n, val_ratio, regions, y, data_selection, name)
465 with open(out_path, 'w') as f:
466 json.dump(samples_dict, f)
--> 468 return MultiArray_1D(x_train_samples), MultiArray_1D(y_train_samples), MultiArray_1D(x_val_samples), MultiArray_1D(y_val_samples)
...
--> 418 assert len(self.array_list) > 0, "Input list is empty. Please provide a list with numpy arrays."
419 assert all(isinstance(item, (np.ndarray, np.memmap)) for item in self.array_list), "Input list should only contain numpy arrays."
421 self.cumulative_sizes = [i for i in range(len(self.array_list))]
AssertionError: Input list is empty. Please provide a list with numpy arrays.