katre / bibliodream

my NaNoGenMo 2016 entry

Bibliophile's Dream

National Novel Generation Month has kicked off, and this year I'm going to take a stab. Based on Google's Deep Dream, I'm going to train a neural net on books (probably from Gutenberg, but we'll see), then run it backwards and see what gets generated.

Step 0: Project Setup

Download and install TensorFlow

Install the gutenberg pip module:

pip install gutenberg

Step 1: Obtaining Texts

Download from Gutenberg, with category data for each.

Probably need to trim off the legal boilerplate.

Step 2: Train the Net

I'll base this on TensorFlow.

Step 3: Run It Backwards

Base this on the code samples at Deep Dream.