0.2.0 (12-05-2018)
------------------
- Considerably faster training
- Smaller models because data that doesn't affect the result is removed during training
- Model saving and loading functions take AsRef<Path> for convenience
- Quotes, dashes and parentheses are now considered word separators as well
- Truecasing sentence fragments (phrases) is now supported
- First letters of sentences are experimentally assumed to be capitalized
- CLI tool now properly reports errors instead of panicing
- CLI tool accepts stding for input for training
0.1.0 (13-01-2018)
------------------
Initial release.