Changed branch name and Pull Request Title to accomodate implementations for jax and tensorflow.
Implemented INNs for pytorch, tensorflow and jax.stax. User can concatenate multiple coupling layers to compose an INN.
User can set backward = True and compute input feature map with the existing network from the output feature map.
Drawback : input and output feature map shapes for each coupling layer should be identical (bijectivity is important for inversion).
Python naming conventions were satisfied according to comment made previously.
Changed branch name and Pull Request Title to accomodate implementations for jax and tensorflow.
Implemented INNs for pytorch, tensorflow and jax.stax. User can concatenate multiple coupling layers to compose an INN. User can set backward = True and compute input feature map with the existing network from the output feature map. Drawback : input and output feature map shapes for each coupling layer should be identical (bijectivity is important for inversion).
Python naming conventions were satisfied according to comment made previously.