---
license: mit
---
This model was created for Cosmocook 2, a project at the 2024 Berkeley AI Hackathon.

Cosmo-demo is a fine-tuned version of the Moondream vision model. This model was trained on a set of images of avocadoes, bananas, doritos, and onions. With special attention to counting the number of each food item in an image.

This model was trained in Intel's Jupyter notebooks using IPEX.