MLA-C01 Question 171
Select 2You are training a machine learning model to classify images on AWS SageMaker. After deploying the model, you notice that the model size is too large for your target environment. Which of the following factors could contribute to the large model size?
- A
The complexity of the model architecture, such as a deep neural network with many layers and parameters.
- B
The size of the dataset used for training the model.
- C
The choice of activation functions used within the model.
- D
The precision level of the data type used for weights and biases, such as floating-point precision.
- E
The hyperparameters chosen during training, such as learning rate and batch size.
Show answer and explanation
Correct answers: A, D
Explanation
Factors such as the complexity of the model architecture and the precision level of the data type used for weights and biases are key contributors to model size. These determine the number of parameters and how much memory each parameter occupies, respectively. Dataset size, activation functions, and hyperparameters do not directly impact the size of the trained model.
- A. Correct.
The complexity of the model architecture directly impacts model size. More layers and parameters typically mean a larger model size.
- B. Incorrect.
The dataset size does not directly influence the model size. While it may affect training time and performance, it does not determine how large the model itself is.
- C. Incorrect.
The choice of activation functions affects the model's functionality and performance but does not have a direct impact on the size of the saved model.
- D. Correct.
The precision level of the data type for weights and biases can significantly affect model size. For example, using float32 results in a larger model compared to using float16 or quantized integers.
- E. Incorrect.
Hyperparameters like learning rate and batch size affect the training process but do not directly influence the size of the final model.