| # HF Space startup hook — runs before app.py. | |
| # | |
| # The default backend is hf_inference (fine-tuned model via HF Inference API). | |
| # No local model servers are needed on HF Spaces. | |
| # | |
| # Local dev with llama.cpp: | |
| # scripts/dev.sh --llamacpp | |
| # Local dev with Modal: | |
| # scripts/dev.sh --modal-llm | |
| set -euo pipefail | |
| source scripts/_env.sh | |
| echo "Startup complete. Using RECEIPT_BACKEND=${RECEIPT_BACKEND:-hf_inference}" | |