Google Professional Machine Learning Engineer Question 34
Select 3Google Cloud PlatformYou are tasked with building a customer support chatbot for a retail company using Google Cloud. The chatbot should understand natural language queries and provide responses based on a knowledge base. Your team has decided to use a pre-trained foundational model with minimal fine-tuning. Which of the following considerations should you focus on while implementing the solution?
- A
Evaluate the model's pre-trained capabilities to ensure it aligns with the business requirements.
- B
Host the foundational model on a local server for better control over the infrastructure.
- C
Ensure the model's input data is pre-processed to match the format the model expects.
- D
Consider data privacy and compliance requirements for user-provided inputs.
- E
Develop custom machine learning models from scratch instead of using pre-trained models.
Show answer and explanation
Correct answers: A, C, D
Explanation
When using foundational models for AI solutions, it is important to evaluate the model's capabilities to ensure alignment with the business problem, properly format the input data, and address any privacy or compliance concerns. These considerations help ensure an efficient and effective deployment while leveraging the benefits of pre-trained models.
- A. Correct.
Evaluating the model's pre-trained capabilities is crucial to confirm that it aligns with the tasks and requirements of the chatbot. This avoids unnecessary development effort.
- B. Incorrect.
Hosting the model on a local server is not necessary, as Google Cloud provides managed services for foundational models, ensuring scalability and reliability.
- C. Correct.
Ensuring the model's input data is in the required format is critical for proper functioning since foundational models expect data in specific structures.
- D. Correct.
Considering data privacy and compliance is essential when dealing with user inputs, especially for a retail chatbot that might handle sensitive information.
- E. Incorrect.
Developing custom models from scratch is not necessary when using pre-trained foundational models, as they are designed to reduce the complexity of implementation.