Revise package structure to ease usability

This is a user-experience-focused package. Without a cogent API,
  it would seem hypocritical and insincere to make another crap
  tool that promises the world.

README.md

@@ -4,4 +4,32 @@ This API is undergoing wild changes as it approaches release.
 * Almost every change will be a breaking change.
 * __Do not__ rely on the functions as they currently are
 
-Please feel free to look around, but bookmark which release tag it was. Master will be changing, viciously.
+Please feel free to look around, but bookmark which release tag it was. Master will be changing, viciously.
+
+
+---
+
+# Understanding [MorphNet](https://arxiv.org/abs/1711.06798)
+
+A Stephen Fox endeavor to become an Applied AI Scientist.
+
+## Setup (to work alongside me)
+
+`git clone https://github.com/stephenjfox/Morph.py.git`
+
+## Requisites
+
+### [Install Anaconda](https://www.anaconda.com/download/)
+* They've made it easier with the years. If you haven't already, please give it a try
+
+### Install Pip
+
+1. `conda install pip`
+2. Proceed as normal
+
+### Dependencies
+
+- Jupyter Notebook
+  * And a few tools to make it better on your local environment like `nb_conda`, `nbconvert`, and `nb_conda_kernels`
+- Python 3.6+ because [Python 2 is dying](https://pythonclock.org/)
+- PyTorch (`conda install torch torchvision`)

demo.py

@@ -1,8 +1,23 @@
-from .my_morph import MorphNet
+import torch
+import torch.nn as nn
+from torch.utils.data import TensorDataset, DataLoader
 
+import morph
+import morph.nn as net
+from morph.layers.sparse import sparsify
+
+from morph._models import EasyMnist
 
 def main():
-    pass
+    my_model = EasyMnist()
+    # do one pass through the algorithm
+    modified = morph.once(my_model)
+
+    my_dataloader = DataLoader(TensorDataset(torch.randn(2, 28, 28)))
+
+    # get back the class that will do work
+    morphed = net.Morph(my_model, epochs=5, dataloader=my_dataloader)
+    morphed.run_training()
 
 
 if __name__ == '__main__':

morph/__init__.py

@@ -1,2 +1 @@
-from .morph import morph
-from .nn.morph_net import *
+from .nn.morph import once # facility tate "morph.once"

morph/_models.py

@@ -0,0 +1,21 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+class EasyMnist(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.linear1 = nn.Linear(784, 1000)
+        self.linear2 = nn.Linear(1000, 30)
+        self.linear3 = nn.Linear(30, 10)
+
+    def forward(self, x_batch: torch.Tensor):
+        """Simple ReLU-based activations through all layers of the DNN.
+        Simple and effectively deep neural network. No frills.
+        """
+        _input = x_batch.view(-1, 784) # shape for our linear1
+        out1 = F.relu(self.linear1(x_batch))
+        out2 = F.relu(self.linear2(out1))
+        out3 = F.relu(self.linear3(out2))
+
+        return out3

morph/layers/__init__.py

@@ -1,2 +1,2 @@
-from .sparsify import *
+from .sparse import *
 from .widen import widen

morph/nn/__init__.py

@@ -0,0 +1,2 @@
+from ._morph_net import Morph
+from .morph import once

morph/nn/_morph_net.py

@@ -0,0 +1,25 @@
+from morph.layers.sparse import percent_waste
+import torch.nn as nn
+from torch.utils.data import DataLoader
+
+class Morph(nn.Module):
+    """An encapsulation of the benefits of MorphNet, namely:
+    1. automatically shrinking and widening, to produce a new architecture w.r.t. layer widths
+    2. Training of the network, to match (or beat) model performance
+    3. 
+    """
+
+    @classmethod
+    def shrink_out(cls, child_layer):
+        new_out = int(child_layer.out_features * percent_waste(child_layer))
+        return nn.Linear(child_layer.in_features, new_out)
+
+    def __init__(self, net: nn.Module, epochs: int, dataloader: DataLoader):
+        super().__init__()
+        self.layers = nn.ModuleList([
+            Morph.shrink_out(c) for c in net.children()
+        ])
+
+    def run_training(self):
+        """Performs the managed training for this instance"""
+        pass

morph/nn/morph.py

@@ -0,0 +1,12 @@
+import torch.nn as nn
+
+def once(net: nn.Module, experimental_support=False) -> nn.Module:
+    """Runs an experimental implementation of the MorphNet algorithm on `net`
+      producing a new network:
+      1. Shrink the layers o
+
+    Returns: either `net` if `experimental_support == False` or a MorphNet of
+        the supplied `net`.
+    """
+    # TODO: run the algorithm
+    return net

setup.py

@@ -5,7 +5,7 @@
     name='morph-py',  # How you named your package folder (MyLib)
     packages=find_packages(),  # Chose the same as "name"
     # Start with a small number and increase it with every change you make
-    version='0.0.5-alpha',
+    version='0.0.6-beta',
     # Chose a license from here: https://help.github.com/articles/licensing-a-repository
     license='GNU',
     # Give a short description about your library
@@ -15,7 +15,7 @@
     # Provide either the link to your github or to your website
     url='https://github.com/stephenjfox/Morph.py',
     # I explain this later on
-    download_url='https://github.com/stephenjfox/Morph.py/archive/v0.0.5.tar.gz',
+    download_url='https://github.com/stephenjfox/Morph.py/archive/v0.0.6.tar.gz',
     keywords=[
         'machine learning', 'deep learning', 'nas', 'architecture',
         'neural networks'