Update app.py

app.py

@@ -54,11 +54,11 @@ def reverse(user_msg: str) -> str:
         )
     prompt = build_prompt(user_msg, tokenizer)
     with torch.inference_mode():
+        encoded = tokenizer(prompt, return_tensors="pt").to(model.device)
         out = model.generate(
-            tokenizer(prompt, return_tensors="pt").to(model.device),
+            **encoded,                # <- pass as keyword args
             max_new_tokens=64,
-            do_sample=False,
-            temperature=0.0
+            do_sample=False           # temperature becomes irrelevant
         )
     full = tokenizer.decode(out[0], skip_special_tokens=True)
     return full[len(prompt):]       # strip the prompt – return only the reply