Update generate.py

generate.py

@@ -228,9 +228,9 @@ def generate(
         **model_kwargs,
     )
 
+    generated_text = self.tokenizer.decode(generated_token_ids[0], skip_special_tokens=True)
+
     if dynamic_temperature is not None:
-        # Decode the generated token IDs into text
-        generated_text = self.tokenizer.decode(generated_token_ids[0], skip_special_tokens=True)
         return generated_text
     else:
-        return generated_token_ids
+        return generated_token_ids, generated_text