Setting up Ivy

Ivy is not a customer facing application so this will take more time in configuration and setting up than getting it running.

Prerequisites

You will need Python 3.6 and Ivy's Core dependencies, mainly:

• Flask
• nltk
• TensorFlow

You will also need the following:

• Git Bash
• Git

To install the dependencies, just run pip install -r requirements.txt, and it will install the requirements automatically.

The git clone doesn't come with ChatLearner, which is a submodule so you need to run git submodule update --init on Ivy's folder.

Setting up

Ivy will require you to "compile" the dataset into a model the program can use.

Keep in mind you'll need a powerful computer to compile the model! We don't recommend at all compiling the model at your laptop or Chromebook, have mercy on your laptop insurance or your parents' wallets.

Compiling the Dataset

Compiling the dataset requires a enormous specification to train the chatbot. However, this can speed up if you use Cloud TPUs for AI training or using a powerful GPU.

to train the chatbot, clone Fern first.

$ git clone https://github.com/ProjectMonika/Fern

Copy Fern to Data inside resources.

$ cp -vRf Fern/Corpus/* resources/corpus
$ cp -vRf Fern/KnowledgeBase/* resources/knowledgebase

Once done, run bash setup.sh, and wait for a while. Training takes some time and computing power so don't use the computer if its running at high I/O.

If the Python script finishes, that means you've successfuly trained the AI!