Update app.py

app.py

@@ -179,7 +179,7 @@ class BaseGenerator(ABC):
         self.cache = ResponseCache(cache_size)
         self.batch_processor = BatchProcessor(max_batch_size)
         self.health_check = HealthCheck()
-        self.tokenizer = self.model_manager.tokenizers[model_name]
+       # self.tokenizer = self.model_manager.tokenizers[model_name]
         #self.tokenizer = self.load_tokenizer(llama_model_name)  # Add this line to initialize the tokenizer
         self.default_config = default_generation_config or GenerationConfig()
         self.model_config = model_config or ModelConfig()
@@ -413,9 +413,24 @@ class LlamaGenerator(BaseGenerator):
         model_config: Optional[ModelConfig] = None,
         cache_size: int = 1000,
         max_batch_size: int = 32,
+        self.tokenizer = self.load_tokenizer(llama_model_name)
 #        self.tokenizer = self.load_tokenizer(llama_model_name)  # Add this line to initialize the tokenizer
 
     ):
+
+        #self.tokenizer = self.load_tokenizer(llama_model_name)  # Add this line to initialize the tokenizer
+
+        def load_model(self, model_name: str):
+            # Code to load your model, e.g., Hugging Face's transformers library
+            from transformers import AutoModelForCausalLM
+            return AutoModelForCausalLM.from_pretrained(model_name)
+    
+        def load_tokenizer(self, model_name: str):
+            # Load the tokenizer associated with the model
+            from transformers import AutoTokenizer
+            return AutoTokenizer.from_pretrained(model_name)
+            
+        
         super().__init__(
             llama_model_name,
             device,