Update app.py

app.py

@@ -5,13 +5,10 @@ import os
 import torch
 import optimum
 from transformers import (AutoModelForCausalLM, AutoTokenizer, LlamaForCausalLM, LlamaTokenizer, GenerationConfig, pipeline,)
-from fastapi.middleware.cors import CORSMiddleware
-
 os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'expandable_segments:True'
 torch.cuda.empty_cache()
 torch.cuda.set_per_process_memory_fraction(0.8)  # Adjust the fraction as needed
 
-app = FastAPI(root_path="/api/v1")
   
 # Load the model and tokenizer
 model_name_or_path = "TheBloke/Wizard-Vicuna-7B-Uncensored-GPTQ"
@@ -39,7 +36,9 @@ def load_model_norm():
 
 model, tokenizer = load_model_norm()
 
-    
+#Now we can init the FlaskApi
+app = FastAPI(root_path="/api/v1")
+
 # Function to generate a response using the model
     
 def generate_response(prompt: str) -> str: