TensorFlow.js' Toxicity Classifier In Action

A demo of TensorFlow.js’ Toxicity Classifier. The Machine Learning model aims at detecting toxic comments posted online, i.e. texts containing insults, threats, attacks, obscenity, etc.

The tool is also available as a WordPress Plugin.

Analysis in progress …
Analysis in progress …
The classifier’s outputs are probabilities, i.e. numbers between 0% and 100%.
Identity Attack
Severe Toxicity
Sexual Explicit