LingSante-MedGuard / scripts /deploy_model.sh
ant-lingfei's picture
Add LingSante MedGuard deployment script
211213e verified
Raw History Blame Contribute Delete
624 Bytes
#!/usr/bin/env bash
# Serve a local model through a vLLM OpenAI-compatible endpoint.
set -euo pipefail
if [[ "${1:-}" == "--help" || $# -lt 1 ]]; then
echo 'Usage: bash deploy_model.sh MODEL [additional vLLM options]'
echo 'Environment: HOST (127.0.0.1), PORT (8000), TP (1), MODEL_NAME (LingSante MedGuard), VLLM_API_KEY (optional)'
exit 0
fi
MODEL="$1"
shift
exec "${PYTHON_BIN:-python3}" -m vllm.entrypoints.openai.api_server \
--model "$MODEL" \
--served-model-name "${MODEL_NAME:-LingSante MedGuard}" \
--host "${HOST:-127.0.0.1}" \
--port "${PORT:-8000}" \
--tensor-parallel-size "${TP:-1}" \
"$@"