NCA-AIIO Question 88
Select 3You are tasked with analyzing the performance of an AI model deployed on an NVIDIA DGX system. During the evaluation, you notice that the model's accuracy fluctuates significantly across different datasets. Which factors should you investigate to identify any relationships or trends affecting the results?
- A
Variations in dataset quality, such as noise or incomplete data
- B
The GPU temperature during model training and inference
- C
Differences in hyperparameter tuning across datasets
- D
The type of AI framework used to develop the model
- E
The batch size used during training and inference
Show answer and explanation
Correct answers: A, C, E
Explanation
Factors such as dataset quality, hyperparameter tuning, and batch size can introduce trends or relationships that affect model performance. Identifying these variables is crucial for ensuring consistent results and understanding the root causes of performance fluctuations. GPU temperature and the choice of AI framework are less likely to directly influence model accuracy in this context.
- A. Correct.
Variations in dataset quality, such as noise or incomplete data, can directly impact the performance of the model by introducing inconsistencies in input data.
- B. Incorrect.
The GPU temperature during model training and inference is unlikely to directly affect the model’s accuracy, as modern GPUs have thermal management systems to maintain stable performance.
- C. Correct.
Differences in hyperparameter tuning across datasets can lead to variations in results because hyperparameters significantly influence training outcomes.
- D. Incorrect.
The type of AI framework used to develop the model generally does not cause fluctuating accuracy on datasets, as most frameworks offer similar core functionalities.
- E. Correct.
The batch size used during training and inference can influence convergence and performance, potentially leading to accuracy variations.