Turi Text Classifier

DEPRECATION WARNING

This code example was intended for use by the legacy Skafos platform and is no longer being maintained. On 05/29/2019, a new version of Skafos was released, streamlining model delivery to the edge.

Sign-up for an account, join our Slack community, and explore some example models to get started.

---

The following repo contains code for training a text classifier model on Skafos using the Turi Create framework. The example model is based on user reviews from Yelp, and given a new sentence, will predict the user's sentiment (negative or positve on a scale from 1-5).

What is here?

The components of this repo are:

• text_classifier.ipynb - a Python notebook that trains and saves a sentiment classifier model to use in your app. Start here.
• utilities/ - a directory that contains helper functions used by text_classifier.ipynb.
• advanced_usage/ - a directory that contains additional information about this text classification model, how to incorporate your own data, advanced usage, and additional example models.
• requirements.txt - a file describing all required Python dependencies.

About the model

• The text classifier is trained on Yelp review data.
• Once trained, you can give the model a snippet of text, and it will predict a sentiment score between 1-5.
  • A score of 1 means that the text is negative in nature.
  • A score of 5 means that the text is positive in nature.
• The model takes about 15 minutes to train in the JupyterLab session on CPUs. To decrease this run time, you can deploy as a job and ask Skafos for more resources. To read more about this, check out Skafos Jobs documentation.

Going beyond the example

• If you wish to incorporate your own data or try another type of text classification model, check out the advanced_usage/ section.
• Turi Create has built-in model evaluation and prediction techniques. We use some of the functions in the advanced_usage/ section, but for a more detailed description, refer to Turi Create's documentation.

Need Help?

Please contact us with questions or feedback! Here are two ways:

• Signup for our Slack Channel
• Find us on Reddit

Also checkout Turi Create's documentation on text classification basics.