20 / #39 added a module models/fully_conv_net.py refactored from models/models1.py, which holds some basic PyTorch neural net architecture. We never started using it, though-- #43 tracked the work to swap from the old to refactored modules.
We should clean up related code if we're to use it. The "mixin" superclasses and final_transformation code are wonky, and should be simplified.
20 / #39 added a module
models/fully_conv_net.py
refactored frommodels/models1.py
, which holds some basic PyTorch neural net architecture. We never started using it, though-- #43 tracked the work to swap from the old to refactored modules.We should clean up related code if we're to use it. The "mixin" superclasses and
final_transformation
code are wonky, and should be simplified.Supplants #40 , #43 .