[NLU] How to compare NLU Model utterance testing between DEV and TEST instances and get the same prediction results in NLU Workbench or NLU Batch Testing, when making continuous development changes to the NLU Model on DEVSummaryWhen testing an NLU Model with test utterances and comparing the prediction results on different instances, it is essential that you compare the exact same NLU Model, as any changes to it, can impact the prediction results. Therefore, if you have a continuous development model, in order to compare NLU models across instances, you must ensure that there are no differences i.e. Development should be stopped during (and leading up to) the comparison. This can be a challenge when you are making continuous changes to the NLU Model on the DEV instance, but only infrequently transfer the NLU Model to the TEST instance for testing the NLU Model.InstructionsIn order to ensure changes to the NLU Model on DEV are not impacting the comparison testing of the same NLU Model across different instances, when you freeze development to transfer the NLU Model to the TEST instance, you should also clone the NLU Model on the DEV instance, so that the cloned NLU Model can be used for the NLU Model testing between DEV and the other instances where the same version NLU Model was transferred to. This will ensure that the NLU Model between instances are the same ensuring the same prediction results and enables you to continue with your NLU Model development on the original non-cloned NLU Model on DEV. Step 1: Clone the NLU Model during the development freeze on DEV. In the NLU Workbench, the Clone menu option is displayed by clicking on the 3 dots at the end of the row for each NLU Model, as follows - Step 2: Provide NLU Model clone details and date of transfer to TEST or other instances during the development freeze on DEV. For example, in the Name and Description fields provide all the NLU Model details for the cloned version, so that on DEV this NLU Model is recognised as a cloned NLU Model to be used for testing purposes only, as follows - Step 3: Once cloned, you then need to train the cloned NLU Model. Once trained, you can start testing utterances in the NLU Workbench or use NLU Batch Testing on this trained cloned NLU Model to compare the prediction results on the same version NLU Model that was transferred to TEST or any other instances. This ensures you can continue development on the original non-cloned NLU Model, whilst testing the clone version against the same NLU Model on other instances.