handler v7

handler.py

@@ -86,18 +86,19 @@ class EndpointHandler:
                     **inputs,
                     **gen_params
                 )
-            logger.debug(f"Output shape: {output_tokens.shape}")
+            logger.info(f"Output shape: {output_tokens.shape}")
             
             # Decode the response
-            logger.debug("Decoding response")
+            logger.info("Decoding response")
             output_text = self.tokenizer.batch_decode(output_tokens)[0]
             
             # Extract the assistant's response by removing the input prompt
             response = output_text[len(prompt):].strip()
             logger.info(f"Generated response length: {len(response)}")
-            logger.debug(f"Generated response: {response}")
+            logger.info(f"Generated response: {response}")
             
-            return [{"role": "assistant", "content": response}]
+            #return [{"role": "assistant", "content": response}]
+            return response
             
         except Exception as e:
             logger.error(f"Error during generation: {str(e)}", exc_info=True)