Command line tool for improving typing speed and accuracy. The main goal is to help programmers practise programming languages.

Demo

Installation

Python environment

pip install --upgrade mltype

Docker

Make sure that Docker and Docker Compose are installed.

docker-compose run --rm mltype

You will get a shell in a running container and the mlt command should be available.

See the documentation for more information.

Main features

Text generation

Using neural networks to generate text. One can use pretrained networks (see below) or train new ones from scratch.
Alternatively, one can read text from a file or provide it manually

Typing interface

Dead simple (implemented in curses)
Basic statistics - WPM and accuracy
Setting target speed
Playing against past performances

Documentation and usage

Detailed documentation: https://mltype.readthedocs.io/en/latest/index.html.
GIF examples: https://mltype.readthedocs.io/en/latest/source/examples.html.

The entrypoint is mlt. To get information on how to use the subcommands use the --help flag (e.g. mlt file --help).

$ mlt
Usage: mlt [OPTIONS] COMMAND [ARGS]...

  Tool for improving typing speed and accuracy

Options:
  --help  Show this message and exit.

Commands:
  file    Type text from a file
  ls      List all language models
  random  Sample characters randomly from a vocabulary
  raw     Provide text manually
  replay  Compete against a past performance
  sample  Sample text from a language
  train   Train a language

Pretrained models

See below for a list of pretrained models. They are stored on a google drive and one needs to download the entire archive.

Name	Info	Link
C++	Trained on https://github.com/TheAlgorithms/C-Plus-Plus	link
C#	Trained on https://github.com/TheAlgorithms/C-Sharp	link
CPython	Trained on https://github.com/python/cpython/tree/master/Python	link
Crime and Punishment	Trained on http://www.gutenberg.org/ebooks/2554	link
Dracula	Trained on http://www.gutenberg.org/ebooks/345	link
Elixir	Trained on https://github.com/phoenixframework/phoenix	link
Go	Trained on https://github.com/TheAlgorithms/Go	link
Haskell	Trained on https://github.com/jgm/pandoc	link
Java	Trained on https://github.com/TheAlgorithms/Java	link
JavaScript	Trained on https://github.com/trekhleb/javascript-algorithms	link
Kotlin	Trained on https://github.com/square/leakcanary	link
Lua	Trained on https://github.com/nmap/nmap	link
Perl	Trained on https://github.com/mojolicious/mojo	link
PHP	Trained on https://github.com/symfony/symfony	link
Python	Trained on https://github.com/TheAlgorithms/Python	link
R	Trained on https://github.com/tidyverse/ggplot2	link
Ruby	Trained on https://github.com/jekyll/jekyll	link
Rust	Trained on https://github.com/rust-lang/rust/tree/master/compiler	link
Scala	Trained on https://github.com/apache/spark/tree/master/mllib	link
Scikit-learn	Trained on https://github.com/scikit-learn/scikit-learn	link
Swift	Trained on https://github.com/raywenderlich/swift-algorithm-club	link

Once you download the file, you will need to place it in ~/.mltype/languages. Note that if the folder does not exist you will have to create it. The file name can be changed to whatever you like. This name will then be used to refer to the model.

To verify that the model was downloaded succesfully, try to sample from it. Note that this might take 20+ seconds the first time around.

mlt sample my_new_model

Feel free to create an issue if you want me to train a model for you. Note that you can also do it yourself easily by reading the documentation (mlt train) and getting a GPU on Google Colab (click the badge below for a ready to use notebook).

Credits

This project is very much motivated by the The Unreasonable Effectiveness of Recurrent Neural Networks by Andrej Karpathy.

jankrepl / mltype

readme