nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #144

You have recently created a proof-of-concept (POC) deep learning model. You are satisfied with the overall architecture, but you need to determine the value for a couple of hyperparameters. You want t

The correct answer is B. Use UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning. For hyperparameter tuning on Vertex AI, integer hyperparameters like embedding dimension typically use UNIT_LINEAR_SCALE, while floating-point hyperparameters like learning rate, which often span orders of magnitude, should use UNIT_LOG_SCALE.

Submitted by diego_uy· Apr 18, 2026ML model development

Question

You have recently created a proof-of-concept (POC) deep learning model. You are satisfied with the overall architecture, but you need to determine the value for a couple of hyperparameters. You want to perform hyperparameter tuning on Vertex AI to determine both the appropriate embedding dimension for a categorical feature used by your model and the optimal learning rate. You configure the following settings: - For the embedding dimension, you set the type to INTEGER with a minValue of 16 and maxValue of 64. - For the learning rate, you set the type to DOUBLE with a minValue of 10e-05 and maxValue of 10e-02. You are using the default Bayesian optimization tuning algorithm, and you want to maximize model accuracy. Training time is not a concern. How should you set the hyperparameter scaling for each hyperparameter and the maxParallelTrials?

Options

  • AUse UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning
  • BUse UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning
  • CUse UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning
  • DUse UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning

How the community answered

(40 responses)
  • A
    3% (1)
  • B
    80% (32)
  • C
    5% (2)
  • D
    13% (5)

Why each option

For hyperparameter tuning on Vertex AI, integer hyperparameters like embedding dimension typically use `UNIT_LINEAR_SCALE`, while floating-point hyperparameters like learning rate, which often span orders of magnitude, should use `UNIT_LOG_SCALE`.

AUse UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning

This choice is incomplete in the prompt but would be incorrect if it suggests UNIT_LINEAR_SCALE for learning rate.

BUse UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learningCorrect

The embedding dimension is an integer hyperparameter that represents a linear range (16 to 64), making `UNIT_LINEAR_SCALE` appropriate for uniform sampling across this range. The learning rate (10e-05 to 10e-02) is a floating-point hyperparameter that typically performs best when sampled from a logarithmic scale, as changes of an order of magnitude are more significant than linear changes in its small range, thus `UNIT_LOG_SCALE` is the correct choice.

CUse UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning

`UNIT_LOG_SCALE` is generally not suitable for embedding dimensions as they are integer values often explored linearly; `UNIT_LINEAR_SCALE` is incorrect for the learning rate which should be sampled logarithmically.

DUse UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning

`UNIT_LOG_SCALE` is generally not suitable for embedding dimensions as they are integer values often explored linearly; `UNIT_LINEAR_SCALE` is incorrect for the learning rate which should be sampled logarithmically.

Concept tested: Vertex AI hyperparameter tuning scaling types

Source: https://cloud.google.com/vertex-ai/docs/training/hyperparameter-tuning-overview#scaling-types

Topics

#Hyperparameter tuning#Vertex AI#Deep Learning#Bayesian Optimization

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice