Update app.py

app.py

@@ -28,10 +28,6 @@ def init():
     histories = []
     #model = None
     
-    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")
@@ -48,11 +44,10 @@ def init():
 
 @spaces.GPU(duration=120)
 def generate_text(messages):
-#    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)
+    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)
     result = text_generator(messages, max_new_tokens=256, do_sample=True, temperature=0.7)
 
     generated_output = result[0]["generated_text"]