Before we start, we need to choose an order. This is the number of words that are taken into account when choosing the next one. Effectively, a larger number produces a more coherent ramble, but might just end up repeating the training text. Usually stick with 2 unless you're adding a very long training text, in which case try turining it up.
Next, you need to enter some text on which to train the Markov chain. You can do this as many times as you like.
Optionally, you can adjust the weightings of your source material.
| Chain | Training lines | Connections | Weight | Weighted Training lines | Weighted Connections |
|---|
Lastly, generate some text. If you enter text, it will be used as a starting point, but only if the last few words appear in the training text.