Latency Troubleshooting Quiz
5 questions Pass: 70% +25 pts (50 with Pro)
Quiz covering Troubleshooting ML
What you'll be asked
- When troubleshooting a sudden spike in model inference latency, which of the following is the most immediate first step to isolate the issue?
- You observe that inference latency is high only for a specific subset of requests. Which tool is best suited for identifying if the bottleneck is in the pre-processing pipeline or the model forward pass?
- What is a common symptom that indicates your model inference latency is being caused by 'data bloat' or excessively large input payloads?
- If your model inference latency follows a 'sawtooth' pattern that correlates with garbage collection cycles, what is the most likely cause?
- You have a model deployed on a GPU. The model throughput is low, but the GPU utilization is consistently at 100%. What is the most likely bottleneck, and how should you address it?
Answers and explanations come after you submit.
Course Content
- Data Cleaning Techniques
- Data Cleaning Techniques Quiz5q
- Feature Scaling Normalization
- Feature Scaling Normalization Quiz5q
- Encoding Techniques
- Encoding Techniques Quiz5q
- Glue DataBrew
- Glue DataBrew Quiz5q
- SageMaker Feature Store
- SageMaker Feature Store Quiz5q
- Ground Truth Labeling
- Ground Truth Labeling Quiz5q
- SageMaker Training Jobs
- SageMaker Training Jobs Quiz5q
- Hyperparameter Tuning
- Hyperparameter Tuning Quiz5q
- Distributed Training
- Distributed Training Quiz5q
- Fine-Tuning Models
- Fine-Tuning Models Quiz5q
- Regularization Techniques
- Regularization Techniques Quiz5q
- Model Registry
- Model Registry Quiz5q
- Ensemble Methods
- Ensemble Methods Quiz5q