Spaces:
Sleeping
Sleeping
OmPranithSai
Configure Render free tier: quantization, build-time model download, mmap loading
98dcd01 | services: | |
| - type: web | |
| name: deepfake-detector | |
| runtime: python | |
| plan: free # NOTE: 512 MB RAM — quantization enabled below to squeeze in | |
| buildCommand: pip install --upgrade pip && pip install -r requirements.txt && python download_model.py | |
| startCommand: gunicorn app:app --workers 1 --threads 4 --timeout 300 | |
| healthCheckPath: /health | |
| envVars: | |
| - key: PYTHON_VERSION | |
| value: "3.11.9" | |
| - key: FLASK_SECRET_KEY | |
| generateValue: true | |
| - key: DEEPFAKE_MAX_FRAMES | |
| value: "24" | |
| # int8 dynamic quantization: required to fit the ViT in 512 MB free tier | |
| - key: DEEPFAKE_QUANTIZE | |
| value: "1" | |
| # Optional: persist the model + SQLite history across deploys with a | |
| # Render persistent disk (requires a plan that supports disks). | |
| # - key: DEEPFAKE_MODEL_DIR | |
| # value: /var/data/models | |
| # disk: | |
| # name: model-disk | |
| # mountPath: /var/data | |
| # sizeGB: 2 | |