Update app.py

app.py

@@ -17,41 +17,39 @@ if not huggingface_token:
     print("no HUGGINGFACE_TOKEN if you need set secret ")
     #raise ValueError("HUGGINGFACE_TOKEN environment variable is not set")
         
-def init():
-    global text_generator
-    
-    
-    
-    #model_id = "microsoft/Phi-3-mini-128k-instruct"
-    
-    
-    
-    tokenizer = AutoTokenizer.from_pretrained(model_id, token=huggingface_token)
-    
-    print(model_id,device,dtype)
-    histories = []
-    #model = None
 
     
     
-    if not is_hugging_face:
-        model = AutoModelForCausalLM.from_pretrained(
-            model_id, token=huggingface_token ,torch_dtype=dtype,device_map=device
-        )
-        text_generator = pipeline("text-generation", model=model, tokenizer=tokenizer,torch_dtype=dtype,device_map=device ) #pipeline has not to(device)
-        
-        if next(model.parameters()).is_cuda:
-            print("The model is on a GPU")
-        else:
-            print("The model is on a CPU")
     
-        #print(f"text_generator.device='{text_generator.device}")
-        if str(text_generator.device).strip() == 'cuda':
-            print("The pipeline is using a GPU")
-        else:
-            print("The pipeline is using a CPU")
+
+
+
+tokenizer = AutoTokenizer.from_pretrained(model_id, token=huggingface_token)
+
+print(model_id,device,dtype)
+histories = []
+#model = None
+
+
+
+if not is_hugging_face:
+    model = AutoModelForCausalLM.from_pretrained(
+        model_id, token=huggingface_token ,torch_dtype=dtype,device_map=device
+    )
+    text_generator = pipeline("text-generation", model=model, tokenizer=tokenizer,torch_dtype=dtype,device_map=device ) #pipeline has not to(device)
     
-    print("initialized")
+    if next(model.parameters()).is_cuda:
+        print("The model is on a GPU")
+    else:
+        print("The model is on a CPU")
+
+    #print(f"text_generator.device='{text_generator.device}")
+    if str(text_generator.device).strip() == 'cuda':
+        print("The pipeline is using a GPU")
+    else:
+        print("The pipeline is using a CPU")
+
+print("initialized")
 
 @spaces.GPU(duration=120)
 def generate_text(messages):
@@ -90,7 +88,6 @@ def call_generate_text(message, history):
     return ""
 
 demo = gr.ChatInterface(call_generate_text,type="messages")
-init()
+
 if __name__ == "__main__":
-    
     demo.launch(share=True)