Skip to content

mlkit

mlkit is a machine learning toolkit providing components for diffusion models, flow matching, pipelines, datasets, and more.

Quick Start

import mlkit

Hydra examples use run_experiment with a TrainSpec (training) or an InferSpec (inference); see the guides below before diving into the API.

Guides

Recommended reading order:

  1. Running an experiment — Hydra config, run_experiment, TrainSpec, InferSpec, custom tasks, artifacts paths
  2. Experiment tracking — MLflow, WandB, resume via report_run.json
  3. Training loop — epochs vs steps, checkpoints, callbacks, halt
  4. Learning rate — scheduler length under Accelerate

Contributors: Docstring style for API docstrings and the autoDocstring template.

API Reference

  • API Reference — Auto-generated module documentation (experiment and report have dedicated pages)