GitHub Pages · browser inference

🛡️ Chat Moderation Demo

Compare a fast TF-IDF baseline with a fine-tuned BERT-Tiny model.

Runs locally in your browser. Your message is not sent to an inference API. This is an educational research demonstration, not a clinical crisis-detection or production moderation service. Do not enter real personal or crisis information.

Initialization

Loading both models

TF-IDF + Logistic Regression Waiting to load…
BERT-Tiny Transformer Waiting to load…

Try both models

Enter text to classify

ONNX Runtime Web · WASM CPU

Try an example: