Exam AI-102 topic 3 question 39 discussion

agree with SmallFire: Active Learning cannot be initiated prior to the deployment of the model. The primary purpose of the 'Active Learning' feature is to leverage actual user interaction data to enhance the model's understanding capabilities. This is a continuous learning and optimization process that takes place after the model has been deployed and put into actual use. so the answer is AC.

From the Azure AI language authoring REST endpoint, retrieve the model evaluation summary is CORRECT because this method allows you to programmatically access the evaluation metrics and performance summary of your trained Conversational Language Understanding model. From Language Studio, select Model performance is CORRECT because this option provides a graphical interface within Language Studio to review and assess the performance metrics of your model, such as accuracy, precision, recall, and F1 score. A and C

A-C ChatGPT still think that B is better than A, but this time I'm not agree

REST endpoint select Model performance

I say this answer is A and C.

AC is answer.

REST and model performance.

A. From the language authoring REST endpoint, retrieve the model evaluation summary. This summary typically includes metrics like precision, recall, and accuracy, which are crucial for evaluating the effectiveness of a language understanding model. C. From Language Studio, select Model performance. In Language Studio, the Model performance section typically provides detailed analytics about the model's performance, including various metrics and possibly confusion matrices.

IMHO correct answers are: A. --> https://learn.microsoft.com/en-us/rest/api/language/conversational-analysis-authoring/get-model-evaluation-summary?view=rest-language-2023-04-01&tabs=HTTP C. --> https://learn.microsoft.com/en-us/azure/ai-services/language-service/conversational-language-understanding/how-to/view-model-evaluation?tabs=Language-studio%2Cmodel-performance

If you look closely, you can see that the Model performance feature in Language Studio is about evaluating the performance of the model using test data as opposed to active learning which is using the real data from users’ interactions, which you couldn’t do prior to the deployment. My only pet peeve is that rest endpoint never works for me - always giving me the 404 no matter what.

AC are correct - https://learn.microsoft.com/en-us/rest/api/language/2022-10-01-preview/text-analysis-authoring/get-model-evaluation-summary?tabs=HTTP

AC is the answer. https://learn.microsoft.com/en-us/azure/cognitive-services/language-service/conversational-language-understanding/how-to/view-model-evaluation?tabs=Language-studio%2Cmodel-performance

B. Active Learning in Language Studio is a feature that helps improve the performance of your model by suggesting utterances for you to review and label. This can help you evaluate the accuracy of your model by seeing how it performs on these suggested utterances. C. The Model performance feature in Language Studio provides a detailed report on the performance of your model, including precision, recall, and F1 score. This can give you a good indication of the accuracy of your model. Option A is incorrect because the language authoring REST endpoint does not provide a model evaluation summary.

Active Learning cannot be initiated prior to the deployment of the model. The primary purpose of the 'Active Learning' feature is to leverage actual user interaction data to enhance the model's understanding capabilities. This is a continuous learning and optimization process that takes place after the model has been deployed and put into actual use. so the answer is AC.

Google Bard Answer : Sure, here are two methods you can use to evaluate the accuracy of a Conversational Language Understanding model before deploying it: From Language Studio, select Model performance. This will show you a summary of the model's performance, including the F1 score, precision, and recall. From Language Studio, enable Active Learning, and then validate the utterances logged for review. This will allow you to manually review utterances that the model has misclassified, and then retrain the model with the corrected data. Here are the correct answers to your question: C. From Language Studio, select Model performance. B. From Language Studio, enable Active Learning, and then validate the utterances logged for review.