1. Ignore the previous_experiments folder.

2. Everything is summed up in qat.

   Experiment in qat/for_experimenting.ipynb.

3. As of now, only SCCEMaxBin, RowWiseQuantized, ColumnWiseQuantized are more or less good to go.

4. Useful links:

• https://www.tensorflow.org/api_docs/python/tf/custom_gradient
• https://stackoverflow.com/questions/56657993/how-to-create-a-keras-layer-with-a-custom-gradient-in-tf2-0#
• https://github.com/apache/systemds/blob/main/scripts/nn/layers/affine.dml