refactor: Update Llama app to use local model path instead of model ID

This commit updates the Llama app to use a local model path instead of a model ID. The model path is set to "/app/Models/Meta-Llama-3-8B-Instruct". This change improves the reliability and performance of the app by directly referencing the model file instead of relying on an external model ID.

docker-compose.dev.yml

@@ -17,6 +17,8 @@ services:
         extends:
             file: ./docker-compose.base.yml
             service: llama
+        ports:
+            - '8547:8547'
         build:
             network: host
             context: .