Skip to content

Help

How to find data and models, share your own, learn with the courses, and manage your account.

Getting started

You can browse datasets, models, news and courses without an account. Create a free account to enrol in courses, like datasets and models, and share your own.

Once you are signed in, your Dashboard — in the menu under your initials, top right — brings together your datasets, your models and your courses.

Datasets and models

Every dataset and model has an address made of its owner and its name, such as glitchdata/wine-quality. Its page has three tabs:

  • Card — what it is, where it came from, how to use it, its licence and how to cite it.
  • Data viewer — for tables, the first rows of each CSV file, so you can see what's inside before downloading.
  • Files — every file with its size and where it is hosted.

glitchdata doesn't store the data itself. Downloads go straight to the publisher that hosts each file; we count them on the way. Always check the licence on the card before you reuse anything — most are open, but some are for non-commercial use only.

Use the filters on the Datasets and Models pages to narrow by tag, licence, format or task, and sort by downloads, likes or recent updates.

Sharing your own

Already on Hugging Face? On the New dataset or New model page, paste its Hugging Face ID (like stanfordnlp/imdb) or address and choose Import. glitchdata brings in its card, licence, tags and key facts, and links every file back to Hugging Face. It arrives as a draft: check it, then submit it for review.

Otherwise, describe it yourself:

  1. Choose + New dataset or + New model from your Dashboard.
  2. Give it a title and a name, then write its card in Markdown: what it contains, where it came from, how to use it and any known limitations.
  3. Add a link to each file where it is hosted — a direct link to the file itself, not a web page about it. Links to CSV and TSV files get a preview in the data viewer.
  4. Tick Submit for review and save.

Every submission is checked by an admin before it goes public, and so is every later change, so a live dataset can't have its files swapped without anyone noticing. Until then only you (and admins) can see it. If an admin asks for changes, you'll see their note on your Dashboard under Needs your attention; edit and save to send it back for review.

Courses

Courses are made of short lessons. The first lesson of every course is a free preview; enrol (it's free) to open the rest. Mark each lesson complete as you finish it — your Dashboard shows your progress and a Continue button that takes you to the next lesson.

Organisations

An organisation is a shared name for a team, company or research group. Its members can publish datasets and models under the organisation's name and edit each other's, and they appear on the organisation's page. Organisations are set up by glitchdata admins — ask one if your team would like one.

Your account

  • Signing in: with your email and password, or with Google where it's offered. If you use Google with the same email as an existing account, the two are linked.
  • Your profile is your public page, listing everything you've shared that is live. Your username is its address and the owner name on your datasets and models.
  • Signing out: open the menu under your initials and choose Sign out.

Still stuck? Read about glitchdata, or check the latest news.