Otter Autograding

Assignments are Otter-graded Jupyter notebooks. You have four ways to grade them; Otter Service (in your browser or as the hosted Standalone service) needs no local installation.

For the exact private solutions repository and any pinned otter-grader version, see your course page, Data 8 or Data 6.

Option 1: Otter Service In-Browser

Otter Service In-Browser is the in-browser version of the Standalone service below: upload the assignment's autograder.zip plus the student notebooks (or a Canvas submissions zip), add a batch per assignment, and download a grades CSV per batch. No login, no server, and nothing leaves your computer.

Option 2: Otter Service Standalone

A web-based grader at grader.datahub.berkeley.edu, with no local installation. You authorize once with your GitHub username, upload the student submissions plus the assignment's autograder.zip (from your course's private solutions repo), and download a CSV of grades. The full, course-specific steps render on each course page.

Option 3: Gradescope

Paid service

Gradescope is a paid service that ties assignments back to Canvas; your institution needs a Gradescope license.

Create the assignment in Canvas and Gradescope, attach the assignment's autograder.zip as the programming autograder, and grades push back to the LMS. See Gradescope's docs on Canvas integration and programming assignments.

Option 4: Grade locally

Full control, but requires local setup. Install otter-grader and Docker, then use otter grade.

Match the pinned version

Some courses pin a specific otter-grader version (for example, Data 6 requires 6.1.6). Install exactly the version listed on your course page, or grades may differ.