fix: try to fix hf model dir

core-model-prediction/Dockerfile

@@ -28,7 +28,7 @@ RUN python -c "from transformers import AutoTokenizer, AutoModelForCausalLM; \
     model.save_pretrained('/app/gemma-2b')"
 
 # Model env
-ENV MODEL_DIR=/app/gemma-2b
+ENV MODEL_DIR=gemma-2b
 
 # Make port 8080 available to the world outside this container
 EXPOSE 8080

core-model-prediction/gemma2b_dependencies.py

@@ -13,9 +13,9 @@ class Gemma2BDependencies:
     def __new__(cls):
         if cls._instance is None:
             cls._instance = super(Gemma2BDependencies, cls).__new__(cls)
-            model_dir = os.getenv("MODEL_DIR", "/app/gemma-2b")
-            cls._instance.tokenizer = AutoTokenizer.from_pretrained(model_dir)
-            cls._instance.model = AutoModelForCausalLM.from_pretrained(model_dir)
+            model_dir = os.getenv("MODEL_DIR", "gemma-2b")
+            cls._instance.tokenizer = AutoTokenizer.from_pretrained(model_dir, local_files_only=True)
+            cls._instance.model = AutoModelForCausalLM.from_pretrained(model_dir, local_files_only=True)
             cls._instance.device = DeviceManager()
             cls._instance.model.to(cls._instance.device)
         return cls._instance