I have a notebook in which I am trying to create an Arabic MNIST notebook from numpy arrays directly!
I am not sure why I am getting a negative loss in validation error and test error.
- Import libraries
- Read data and create numpy arrays.
- Created a new Dataset class named
- Reshaped images (I can feed without reshaping as well but I want to create three channels by copying single channel 3 times)
- In ArabicMNIST Dataset I am normalizing per image (which is wrong but I tried to normalize the entire data first as well)
- In ArabicMNIST Dataset class I created 3 channels and transposed the images in [C, H, W] notation.
- Created a Databunch with my new ArabicMNIST dataset.
learnobject and trained for 3 epochs.
Why sometimes most of the times I am getting a negative loss?
Edit: Moved to normal thread. (Better idea?)