I am trying to make predictions on an unlabeled test set (not validation set). pred_batch(DatasetType.Test) always predicts only first batch. How do I get predictions for all other batches?
path = ‘/home/xxx/mnist_png/’
data = ImageDataBunch.from_folder(path, test=“test”, ds_tfms=get_transforms(), size=28, num_workers=8).normalize()
learn = create_cnn(data, models.resnet34, metrics=error_rate)
#load previously trained weights for predictions
learn.load(‘stage-2’)
preds = learn.pred_batch(DatasetType.Test)