Google Professional Machine Learning Engineer Question 181
Select 3Google Cloud PlatformYou are tasked with building a sentiment analysis solution for customer reviews using Google Cloud. Instead of training a model from scratch, you decide to leverage the Model Garden to accelerate development. Which of the following actions are appropriate when using foundational and open-source models from Model Garden?
- A
Select a pre-trained model for text classification and fine-tune it on your customer review dataset.
- B
Directly deploy the pre-trained model without additional training, assuming it generalizes well to your domain.
- C
Evaluate the model's performance on your specific dataset before deploying it to production.
- D
Modify the architecture of the pre-trained model in Model Garden to fit your specific requirements.
- E
Use the TensorFlow Hub integration to easily access and utilize the pre-trained model.
Show answer and explanation
Correct answers: A, C, E
Explanation
Leveraging foundational and open-source models from Model Garden can significantly accelerate development by using pre-trained models. However, it is critical to fine-tune these models on your domain data if needed, evaluate their performance to ensure they meet your use case requirements, and utilize tools like TensorFlow Hub for integration. Modifying the architecture of pre-trained models is typically unnecessary and may complicate the process.
- A. Correct.
This is correct because foundational models in Model Garden are often pre-trained and can be fine-tuned on domain-specific data to improve performance for a given task.
- B. Incorrect.
This is incorrect because pre-trained models may not generalize well to domain-specific datasets without fine-tuning or evaluation.
- C. Correct.
This is correct because it is a best practice to evaluate any model, including pre-trained ones, on your specific dataset to ensure it meets performance requirements.
- D. Incorrect.
This is incorrect because foundational models in Model Garden are typically designed to be used as-is or fine-tuned, not modified at the architectural level.
- E. Correct.
This is correct because TensorFlow Hub integration allows for seamless access to pre-trained models in Model Garden, making it easy to use them in your workflows.