AI-ordlista
Den kompletta ordlistan över AI
Artifact Registry
Centralized storage system for ML artifacts such as model weights, preprocessed datasets, and configuration files with version management.
DVC
Data Version Control, an open-source tool that extends Git to manage versioning of large datasets and ML models with storage optimization.
MLflow Tracking
Component of MLflow that automatically records parameters, metrics, and artifacts from training runs in a centralized server.
Semantic Versioning for Models
Numbering convention (MAJOR.MINOR.PATCH) applied to models to indicate the extent of changes between versions.
Model Metadata
Structured information associated with a model including hyperparameters, performance metrics, dataset used, and training environment.
Model Lineage
Complete dependency graph showing the origin of data, applied transformations, and code versions that led to a specific model.
Model Promotion
Controlled transition of a model from a lower to a higher environment after validating quality and performance criteria.
Model Tagging
Adding descriptive labels to model versions to facilitate search, filtering, and classification according to business or technical criteria.
Model Provenance
Complete documentation of a model's history including the versions of code, data, and configurations that contributed to its creation.
Model Snapshot
Immutable backup of the complete state of a model at a given time including weights, architecture, and configuration for exact reproduction.
Version Conflict Resolution
Strategies and tools to manage conflicts when multiple concurrent changes are made to the same versioned ML resources.