This was a great post!!!
I was able to adapt it without much trouble. For anyone coming along later, the fastai docs say to use vision_learner rather than cnn_learner (https://docs.fast.ai/vision.learner.html#learner-convenience-functions), also if you’re using the version of the dataset on Kaggle, you’ll have to cast to uint8 like this: .astype(np.uint8)
This was a great post!!!
I was able to adapt it without much trouble. For anyone coming along later, the fastai docs say to use vision_learner rather than cnn_learner (https://docs.fast.ai/vision.learner.html#learner-convenience-functions), also if you’re using the version of the dataset on Kaggle, you’ll have to cast to uint8 like this: .astype(np.uint8)