updated docs

docs/source/conf.py

@@ -83,6 +83,7 @@ extensions = [
     'sphinx.ext.autosummary',
     'sphinx.ext.napoleon',
     'recommonmark',
+    'sphinx.ext.autosectionlabel',
     # 'm2r',
     'nbsphinx',
 ]

docs/source/index.rst

@@ -3,13 +3,13 @@
    You can adapt this file completely to your liking, but it should at least
    contain the root `toctree` directive.
 
-Welcome to PyTorch-Lightning!
+PyTorch-Lightning Documentation
 =============================
 
 .. toctree::
-   :maxdepth: 4
+   :maxdepth: 1
    :name: start
-   :caption: Quick Start
+   :caption: Start Here
 
    new-project
    examples
@@ -17,7 +17,7 @@ Welcome to PyTorch-Lightning!
 .. toctree::
    :maxdepth: 4
    :name: docs
-   :caption: Docs
+   :caption: Python API
 
    documentation
 

docs/source/new-project.rst

@@ -1,13 +1,13 @@
 Quick Start
 ===========
-To start a new project define two files, a LightningModule and a Trainer file.    
-To illustrate Lightning power and simplicity, here's an example of a typical research flow.    
+| To start a new project define two files, a LightningModule and a Trainer file.
+| To illustrate the power of Lightning and its simplicity, here's an example of a typical research flow.
 
 Case 1: BERT
 ------------
 
-Let's say you're working on something like BERT but want to try different ways of training or even different networks.  
-You would define a single LightningModule and use flags to switch between your different ideas.
+| Let's say you're working on something like BERT but want to try different ways of training or even different networks.
+| You would define a single LightningModule and use flags to switch between your different ideas.
 
 .. code-block:: python
 
@@ -66,6 +66,10 @@ Then you could do rapid research by switching between these two and using the sa
 
 **Notice a few things about this flow:**
 
-1. You're writing pure PyTorch... no unnecessary abstractions or new libraries to learn.   
-2. You get free GPU and 16-bit support without writing any of that code in your model.   
-3. You also get all of the capabilities below (without coding or testing yourself).     
+1. You're writing pure PyTorch... no unnecessary abstractions or new libraries to learn.
+2. You get free GPU and 16-bit support without writing any of that code in your model.
+3. You also get all of the capabilities below (without coding or testing yourself).   
+
+- :ref:`Examples & Tutorials`
+- :ref:`Examples & Tutorials`
+- :ref:`Examples & Tutorials`

pytorch_lightning/trainer/__init__.py

@@ -16,4 +16,8 @@ This is the basic use of the trainer:
     trainer = Trainer()
     trainer.fit(model)
 
-"""
+The Trainer holds all the engineering code you might need such as distributing over GPUs or early stopping.
+The LightningTemplate holds the core computations, train, val, test loop, optimizer and dataloaders.
+
+This pattern de-couples the engineering from the science which makes your code reusable and free to run on any hardware.
+"""