How to Find Datasets, Algorithms, and Experiments on OpenML

Updated

OpenML organizes its shared content into three main resource types: datasets, algorithms, and experiments. To browse or search them, go to openml.org and use the site's navigation to reach the corresponding resource lists. Note that the OpenML web app requires JavaScript — if the page appears blank or shows only a "You need to enable JavaScript to run this app" message, enable JavaScript in your browser and reload before you can browse anything.

What Each Resource Type Contains

OpenML describes itself as an open platform for sharing datasets, algorithms, and experiments "to learn how to learn better, together." The three categories map to distinct things you can look up:

Resource What you'll find Typical use
Datasets Shared machine learning datasets with metadata Finding data to train or benchmark on
Algorithms Shared algorithm and workflow definitions Reusing or comparing modeling approaches
Experiments Records of runs linking tasks, algorithms, and results Reviewing how methods performed

Because the platform is built around sharing, each resource is tied to the others — an experiment typically references the dataset and algorithm involved, so starting from any one list can lead you to the others.

Steps to Reach the Resource Lists

  1. Open a browser and go to openml.org.
  2. Confirm JavaScript is enabled. If you see only the "enable JavaScript" notice, turn it on in your browser settings and reload the page.
  3. Use the site's top-level navigation to select the resource type you want — datasets, algorithms, or experiments.
  4. On the resource page, use the search or filter controls to narrow results (for example, by name or by related task).
  5. Open an individual entry to see its details and its links to related datasets, algorithms, or experiments.

Expected result: you land on a browsable, searchable list for the resource type you chose, and each entry opens to a detail page.

Common Snags

  • Blank page or the JavaScript notice: the app is client-rendered, so nothing loads without JavaScript. This is the most likely reason a first visit appears to fail.
  • Wrong starting point: if you search from a dataset page but want experiment records, switch resource type first — the lists are separate.
  • Narrow searches returning nothing: broaden the query or clear filters before concluding a resource isn't there.

Once JavaScript is on and you're on the right resource list, finding datasets, algorithms, and experiments on OpenML is a matter of navigating to the category and searching within it.