Small pet project for learning Tensorflow. The goal is to train a neural net to take a word as input and produce the language as output. I have trained it with approximately 10,000 words per language for 3 languages (English, French, and Mandarin).
The project contains 2 folders:
- data: the training data
- word list scraping: jupyter notebooks for scraping the data off the internet. Note that only French and Mandarin words were scraped. English words came from: https://github.com/first20hours/google-10000-english
The main jupyter notebook is where the actual training and testing is.
- saving neural net
- dropout
- use tensorboard for visualization