https://github.com/kylemcdonald/mueller-unredacter

Generating text completions based on the Mueller report
https://github.com/kylemcdonald/mueller-unredacter

Last synced: about 2 months ago
JSON representation

Generating text completions based on the Mueller report

• Host: GitHub
• URL: https://github.com/kylemcdonald/mueller-unredacter
• Owner: kylemcdonald
• License: mit
• Created: 2019-04-19T02:33:50.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2019-04-19T02:42:52.000Z (almost 6 years ago)
• Last Synced: 2024-10-14T15:11:26.355Z (3 months ago)
• Language: HTML
• Size: 4.64 MB
• Stars: 28
• Watchers: 4
• Forks: 3
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md
  • License: license.md

Awesome Lists containing this project

• awesome-rainmana - kylemcdonald/mueller-unredacter - Generating text completions based on the Mueller report (HTML)

README

        
# Mueller unredacter

![Mueller unredacter](screenshot.png)

A speed project generating text completions with an LSTM trained on the Mueller report. I used [data prepared by Factbase](https://docs.google.com/document/d/1d8G6MlxNX5JDoo2ZR3cH3OYiROWVOFBEZZiCF4p6CEM/edit) with a little extra preprocessing to remove headings and footnotes. It's nowhere near perfect but you can find it at `mueller-report.txt`. Then I trained a network using [this port of char-rnn by Sherjil Ozair](https://github.com/sherjilozair/char-rnn-tensorflow) and loaded it with [ml5.js](https://ml5js.org/docs/CharRNN).

It doesn't work on mobile right now, due to some bugs with either ml5.js or TensorFlow.js.