| title: Recipe Lens Chef | |
| sdk: gradio | |
| python_version: "3.12" | |
| suggested_hardware: t4-small | |
| app_file: app.py | |
| # Recipe Lens Chef | |
| This repository contains a minimal Hugging Face Space for turning food photos | |
| into chef-style recipe ideas and cooking guidance with | |
| [`openbmb/MiniCPM-V-4.6`](https://huggingface.co/openbmb/MiniCPM-V-4.6). | |
| The app is intentionally inference-only. It does not include fine-tuning, | |
| dataset preparation, trainer scripts, adapters, checkpoints, or training | |
| artifacts. | |
| ## Local Development | |
| ```bash | |
| uv sync --python 3.12 | |
| uv run python app.py | |
| ``` | |
| The model ID can be overridden without changing code: | |
| ```bash | |
| MODEL_ID=openbmb/MiniCPM-V-4.6 uv run python app.py | |
| ``` | |
| ## Space Deployment | |
| Deployments are owned by GitHub Actions. Pushes to `main` run | |
| `.github/workflows/deploy-space.yml`, which uploads the checked-out GitHub | |
| commit to the Hugging Face Space. | |
| Required GitHub secret: | |
| ```bash | |
| HF_TOKEN | |
| ``` | |
| Optional GitHub variable: | |
| ```bash | |
| HF_SPACE_ID=binzhango/hf-build-small-hackathon | |
| ``` | |
| The Space reads dependencies from `requirements.txt` and starts `app.py` with | |
| Gradio. Python 3.12 is requested in this README's Space metadata. | |