Portfolio Project

Writerly AI

Browser-only text completion demo hosted on GitHub Pages. Enter a prompt and generate a continuation with an on-device language model. Due to hosting contraints, this demo only uses the CPU for inference, to get better performance, try running the code locally with GPU support using python.

Prompt

CPU based models such as this one may struggle exponentially with total token values above 5.

Max New Tokens Temperature Top-P Repetition Penalty

Model not loaded yet.

Connected Output (Prompt + Completion)