In MLflow, which component is used to manage the lifecycle of a machine learning model across stages such as Staging, Production, and Archived?
Model Lifecycle Management
Question · Easy
When using Spark MLlib for distributed training, which class is used to build a sequence of data transformations and model training steps?
Model Development
In Databricks, what is the primary purpose of the Feature Store?
Tech Exam Prep Podcast
Listen on Spotify
Podcast
Which MLflow function logs a trained model along with its signature and input example for reproducibility?
Question · Medium
When performing hyperparameter tuning in Databricks, which library integrates with SparkML to parallelize trial execution across a cluster?
In the context of MLOps on Databricks, what is the purpose of Databricks Asset Bundles (DABs)?
Mlops
Which Databricks feature detects data drift by comparing statistical distributions of incoming data against a reference dataset?
What is the primary advantage of using Delta Lake tables for ML feature storage compared to standard Parquet files?
In MLflow, which concept groups multiple runs together to organize related experiments?
Which approach should be used to serve a trained ML model as a real-time REST API endpoint on Databricks?
Model Deployment
When implementing distributed training on Databricks, which framework enables data-parallel training of deep learning models across multiple GPUs?
In Databricks, what is the recommended approach for automated model retraining when data drift is detected?
Which SparkML class is used to convert categorical string columns into numeric indices for model training?
What is the purpose of model signatures in MLflow?
Which testing strategy is recommended for validating an ML pipeline before promoting it to production in Databricks?
In Databricks, which feature enables A/B testing by routing a percentage of inference traffic to different model versions?
Question · Hard
Which Spark MLlib evaluation metric is appropriate for assessing a binary classification model's performance across all threshold values?
What is concept drift in the context of production ML systems?
Which Databricks tool provides AutoML capabilities for automatically training and evaluating multiple models on a dataset?
When deploying a custom model to Databricks Model Serving, which MLflow model flavor allows packaging arbitrary Python code with custom predict logic?
In SparkML, which class combines multiple feature columns into a single feature vector for model training?
Which MLOps practice ensures that a model training pipeline can be exactly reproduced with the same data, code, and environment?
Which method should be used to perform batch inference on a large dataset using a registered MLflow model on Databricks?
What is the key difference between data drift and prediction drift in production ML monitoring?
In Databricks, which approach is recommended for implementing feature lookup during real-time model serving?
Which cross-validation approach in SparkML splits data into k folds for model evaluation?
In MLflow Model Registry on Databricks with Unity Catalog, what is the correct model naming format?
Which technique reduces overfitting in a gradient-boosted tree model trained with SparkML?
What is the recommended approach for managing multiple environments (dev, staging, production) for ML workflows on Databricks?
Which SparkML transformer normalizes feature values to have zero mean and unit variance?
In Databricks Model Serving, what happens when a model endpoint receives a request with an input schema that does not match the model's signature?
Which statistical test is commonly used by Lakehouse Monitoring to detect numerical feature drift?
Which MLflow function retrieves a specific model version from the Model Registry for loading?
When building an end-to-end ML pipeline on Databricks, which component handles orchestrating the sequence of data preparation, training, evaluation, and deployment steps?
Which technique addresses class imbalance when training a binary classification model?
In Databricks, what role does Unity Catalog play in ML model governance?
Which method is used to log multiple parameters at once during an MLflow run?
What is the primary benefit of using Spark's pandas UDF (vectorized UDF) for model inference compared to standard Python UDFs?
Which monitoring metric is most critical for detecting that a production model's predictions are becoming less accurate over time?
In SparkML, which class builds a grid of hyperparameter combinations for model tuning?
Which Databricks feature provides serverless compute for model serving endpoints, automatically scaling based on request volume?
What is the purpose of the MLflow model conda.yaml file?
Which SparkML algorithm is suitable for recommendation systems using collaborative filtering?
In the context of ML model governance, what is model lineage?
Which technique helps explain individual predictions from a complex ML model by computing feature importance for each prediction?
What is the recommended strategy for rolling back a model serving endpoint to a previous version if the new version shows degraded performance?
Which Databricks feature enables automatic logging of ML experiment parameters, metrics, and models without explicit logging calls?
Loading more…