When the model is trained for intent detection and slot filling tasks separately, the performance is higher when the model is trained for both the tasks together. I mean for multitasking, there is no performance gain? Any comments or help how to get the performance gain over single-task model would be highly appreciated?
When the model is trained for intent detection and slot filling tasks separately, the performance is higher when the model is trained for both the tasks together. I mean for multitasking, there is no performance gain? Any comments or help how to get the performance gain over single-task model would be highly appreciated?