During the prediction in a running real-time service, error { "erno": "MR.0105", "msg": "Recognition failed","words_result": {}} occurred.
Locate the fault by analyzing the error log on the Logs tab of the real-time service details page.
According to the error log shown in the preceding figure, the prediction failure is caused by the model inference code.
According to the error log, mandatory parameters are missing in the append() method. To rectify the fault, modify the code in the model inference code file customize_service.py to transfer proper parameters to the append() method.