r33drichards@r33drichards · Sep 19Adds image understanding to Laya using a small SmolVLM-256M backbone, so you can ask calibrated yes/no, choice and score questions about a picture plus text.
