printing generated ids

handler.py

@@ -37,6 +37,7 @@ class EndpointHandler():
 
         # Decode the generated ids to text
         # Exclude the input_ids length to get only the new tokens
+        print("Generated IDs:", prediction_ids[0, input_ids.shape[1]:])
         prediction_text = self.tokenizer.decode(prediction_ids[0, input_ids.shape[1]:], skip_special_tokens=True)
         return [{"generated_text": prediction_text}]