FastVLM Screenshot Explainer β CPU-only (no uploads)
A tiny Gradio Space that showcases Apple's FastVLM-0.5B on CPU with a curated image gallery. Pick an image β choose a task (Explain / Extract numbers / Write alt-text / Ask a question) β run.
Local dev
pip install -r requirements.txt python app.py
Notes
- Uses
trust_remote_codeas shown on the model card, including IMAGE_TOKEN_INDEX=-200 and the model's own image processor. - License: apple-amlr (Apple Machine Learning Research) β research & non-commercial only.