google/ml-metadata

For recording and retrieving metadata associated with ML developer and data scientist workflows.

What it solves

ML Metadata (MLMD) provides a way to record and retrieve metadata associated with machine learning developer and data scientist workflows. It helps track the lineage and history of ML experiments and data processing steps.

How it works

MLMD is a library that allows users to store and retrieve metadata. It provides a Python API and a gRPC server for managing this metadata store, enabling the tracking of artifacts and execution history within ML pipelines.

Who it’s for

This tool is designed for ML developers and data scientists who need to maintain a structured record of their experimental workflows and data lineage.

Highlights

  • Metadata Store: A dedicated library for recording and retrieving ML-specific metadata.
  • Python API: Easy integration into existing Python-based ML workflows.
  • Cross-Platform Support: Built and tested on macOS and Ubuntu.
  • gRPC Server: Includes a C++ binary for serving metadata storage needs.

Related

  • Project
    openml/openml-python

    A Python API for OpenML that allows users to download and upload datasets, machine learning experiment results, and curated benchmarking tasks.

  • Project
    evidentlyai/evidently

    An open-source Python framework to evaluate, test, and monitor ML and LLM-powered systems, helping developers detect data drift and ensure output quality.

  • Project
    clearml/clearml

    An open-source MLOps and LLMOps suite that provides experiment tracking, orchestration, and data management to streamline the deep learning development lifecycle.

  • Project
    simonw/llm

    A CLI tool and Python library for interacting with various remote and local Large Language Models, featuring SQLite logging and structured data extraction.

  • Project
    aimhubio/aim

    An open-source, self-hosted ML experiment tracker that allows users to log, visualize, and programmatically query metadata for thousands of training runs.