MLS-C01 · Question #319
An ecommerce company has used Amazon SageMaker to deploy a factorization machines (FM) model to suggest products for customers. The company's data science team has developed two new models by using…
The correct answer is D. Create two production variants for the TensorFlow and PyTorch models. Specify the weight for. SageMaker supports multiple production variants on a single endpoint. You can assign traffic weights - e.g., 70% to the FM model, 15% to TensorFlow, 15% to PyTorch - directly on the endpoint configuration. For the European customer routing requirement, SageMaker's inference…
Question
An ecommerce company has used Amazon SageMaker to deploy a factorization machines (FM) model to suggest products for customers. The company's data science team has developed two new models by using the TensorFlow and PyTorch deep learning frameworks. The company needs to use A/B testing to evaluate the new models against the deployed model. The required A/B testing setup is as follows:
- Send 70% of traffic to the FM model, 15% of traffic to the TensorFlow
model, and 15% of traffic to the PyTorch model.
- For customers who are from Europe, send all traffic to the TensorFlow
model. Which architecture can the company use to implement the required A/B testing setup?
Options
- ACreate two new SageMaker endpoints for the TensorFlow and PyTorch models in addition to the
- BCreate two production variants for the TensorFlow and PyTorch models. Create an auto scaling
- CCreate two new SageMaker endpoints for the TensorFlow and PyTorch models in addition to the
- DCreate two production variants for the TensorFlow and PyTorch models. Specify the weight for
How the community answered
(40 responses)- A8% (3)
- B13% (5)
- C3% (1)
- D78% (31)
Explanation
SageMaker supports multiple production variants on a single endpoint. You can assign traffic weights - e.g., 70% to the FM model, 15% to TensorFlow, 15% to PyTorch - directly on the endpoint configuration. For the European customer routing requirement, SageMaker's inference components and context-based routing allow you to override the default traffic split based on request metadata (such as customer geography), sending all European traffic to the TensorFlow variant. Option D describes this correctly: creating two production variants with explicit weights and leveraging per-request routing overrides. Options A and C propose separate endpoints for each model, which would require client-side routing logic and cannot natively handle the 'all European traffic to one model' requirement within a unified A/B test. Option B incorrectly conflates auto scaling with traffic splitting, which are separate concerns.
Topics
Community Discussion
No community discussion yet for this question.