Completion requirements
1. Optimiza la generación de modelos analíticos
Optimizes the generation of analytical models
Description of functionality
MLProject is the main asset in the production of a machine learning model. This asset has been modified by adding two key functionalities to increase its execution capacity: running several workouts of the same version of the model simultaneously (multiRun) and, when required, storing more than one model per training (multiModel). The goal is to efficient data scientists by improving the throughput of the MLOps cycle.
Improvements
- multiRun: Allows you to create multiple MLflow runs in the same execution of a MLProject.
- multiModel: Registration of several models in the same execution of a MLProject, being able to register several models in the same run or in different runs.
- Model tagging: Support for recording tags and parameters in runs of an execution to classify and identify models in an agile way.
- Executions Comparator: The result screen of an execution allows comparisons to be made in a configurable table or using custom graphics (see Insights).
Figure 1:MLProject type artefact for the generation of models.
Figure 2:Comparison of models.