tensorrtllm [0.16] protobuf input data type mismatch

Triton image: nvcr.io/nvidia/tritonserver:24.12-trtllm-python-py3
Triton version: 0.16

Model checkpoint conversion and TRT build creates the converted checkpoint and output engine successfully.

Upon launching Triton, getting this error from TRT LLM
python3 /var/run/models/tensorrtllm_backend/scripts/launch_triton_server.py --model_repo=/var/run/models/tensorrtllm_backend/triton_model_repo --world_size=16

[libprotobuf ERROR /tmp/tritonbuild/tritonserver/build/_deps/repo-third-party-build/grpc-repo/src/grpc/third_party/protobuf/src/google/protobuf/text_format.cc:337] Error parsing text-format inference.ModelConfig: 51:16: Expected integer or identifier, got: $ E0130 21:39:24.547441 1763 model_repository_manager.cc:1460] "Poll failed for model directory 'tensorrt_llm': failed to read text proto from /var/run/models/tensorrtllm_backend/triton_model_repo/tensorrt_llm/config.pbtxt"

Protobuf error source could be below section from file: /var/run/models/tensorrtllm_backend/triton_model_repo/tensorrt_llm/config.pbtxt

{ name: "encoder_input_features" data_type: **${encoder_input_features_data_type}** no data type dims: [ -1, -1 ] allow_ragged_batch: true optional: true },

Start with the generated tensorrt_llm/config.pbtxt and the launch_triton_server.py command shown in the report. Inspect how encoder_input_features_data_type is substituted before Triton parses the model configuration, then reproduce the launch failure with the provided Triton image and model repository. Done means Triton accepts the config without the protobuf parsing error.