Update app.py

app.py

@@ -26,32 +26,28 @@ def load_pipelines():
         "branch service", "transaction delay", "account closure", "information error"
     ]
     
-    device = 0 if torch.cuda.is_available() else -1
-    dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+    dtype = torch.float32  
     
     topic_classifier = pipeline(
         "zero-shot-classification",
         model="MoritzLaurer/deberta-v3-base-zeroshot-v1",
-        device=device,
-        torch_dtype=dtype
     )
     
     # Sentiment Analysis Model
     sentiment_classifier = pipeline(
         "sentiment-analysis",
-        model="cardiffnlp/twitter-roberta-base-sentiment-latest",  # Best performing model from notebook
-        device=device
+        model="cardiffnlp/twitter-roberta-base-sentiment-latest",  
     )
     
     # Reply Generation Model
-    model_name = "Leo66277/finetuned-tinyllama-customer-replies"  # Using the fine-tuned model in the notebook
+    model_name = "Leo66277/finetuned-tinyllama-customer-replies"
     tokenizer = AutoTokenizer.from_pretrained(model_name)
     model = AutoModelForCausalLM.from_pretrained(model_name)
-    model.to(device)
+
     
     def generate_reply(text):
         prompt_text = f"Please write a short, polite English customer service reply to the following customer comment:\n{text}"
-        inputs = tokenizer(prompt_text, return_tensors="pt", truncation=True, max_length=512).to(device)
+        inputs = tokenizer(prompt_text, return_tensors="pt", truncation=True, max_length=512)
         
         with torch.no_grad():
             gen_ids = model.generate(