When I start the server service using the following docker compose file, an error occurs

docker-compose file, os/service/Llamapp_derver_cuda: latest is the image I compiled myself :
version: '3'
services:
llamacpp-server:
image: os/service/llamacpp_server_cuda:latest
container_name: llama-container
ports:
- 8080:8080
volumes:
- /home/gemma3:/models
entrypoint: /app/llama-server
command: >
--model /models/gemma-3-12b-it-q4_0.gguf
--mmproj /models/mmproj-model-f16-12B.gguf
-c 4096
-n 32768
-np 2
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
interval: 30s
timeout: 5s

deploy:
  resources:
    reservations:
      devices:
        - driver: nvidia
          count: all
          capabilities:
            - gpu

error：error: invalid argument: --mmproj
when I use ./build/bin/llama-gemma3-cli -m {text_model}.gguf --mmproj mmproj.gguf --image your_image.jpg ,it can be run
Does llama_server not support gemma3 multimodality, or did I make a mistake