lmi-container-def.json

{
  "Image": "763104351884.dkr.ecr.us-east-1.amazonaws.com/djl-inference:0.36.0-lmi27.0.0-cu130-v1.1",
  "Mode": "SingleModel",
  "ModelDataUrl": "s3://medgemma-async-2026/models/medgemma-vllm/model.tar.gz",
  "Environment": {
    "OPTION_ROLLING_BATCH": "vllm",
    "OPTION_TENSOR_PARALLEL_DEGREE": "1",
    "OPTION_MAX_MODEL_LEN": "4096",
    "OPTION_DTYPE": "auto",
    "OPTION_TRUST_REMOTE_CODE": "true"
  }
}