Skip to content
Simpcat Simpcat Simpcat

Overview

Simpcat is an ML experiment manager. Use it to track runs, stream metrics and media from your training jobs, compare results in the web UI, and inspect run data from Python or the CLI.

Key features

Section titled “Key features”
  • Fast experiment tracking — Start a run with simpcat.init(), log metrics with simpcat.log(), and see charts update in the web UI while training is still running.
  • Rich metrics and media — Track scalars, histograms, images, and audio in the same run so you can inspect both model quality and training behavior in one place.
  • Run configs, summaries, and comparisons — Store hyperparameters alongside metrics, surface summary values in run tables, and compare runs across a project.
  • Reliable background reporting — Every run gets its own reporter process, which batches uploads and uses a WAL-backed local queue for crash recovery and offline sync.
  • Resume and branch experiments — Resume interrupted runs or fork from a previous checkpoint when you want to continue training from an existing state.
  • Python, CLI, and Lightning support — Use the Python SDK directly, query data with the simpcat CLI and API, or plug in SimpCatLogger as a Lightning logger.

Next steps

Section titled “Next steps”
  • Quickstart — install the SDK, save your API key, and log your first run
  • CLI guide — authentication, offline sync, and data queries