Rubric Alignment Checker

How Rubric Alignment Checker works

Press Load model once: your browser downloads Universal Sentence Encoder (MediaPipe) (13 MB) from this site and keeps it in its cache. Then add your text and press Check locally. The model runs on this device, so nothing you enter is sent to a server. Until the model is loaded, or if your device cannot run it, the tool falls back to a simple keyword check and labels the result as the fallback.

The tool splits "Rubric (one criterion per line)" into sentences and list items (each one a rubric item), then uses Universal Sentence Encoder (MediaPipe) to find the closest sentence in the other text by meaning, not by exact words. "Led accessibility audits" can match "accessibility expertise" even though the wording differs.

A rubric item counts as covered when its best match scores 78 or more on a 0-100 scale (cosine similarity rescaled from 0.6); 65-77 is shown as partial. Coverage is the share of rubric items that are covered, and the ones without a close match are listed so you can fix them.

Worked example

The model's output for the built-in sample, recorded on a desktop browser. Your run may word things slightly differently.

Write a two-page argument about renewable energy tradeoffs. State a clear thesis in your first paragraph. Support each point with evidence from at least three cited sources. Address one counterargument and respond to it. End with a conclusion that explains the policy impact. Clear thesis statement. Uses quality evidence from credible sources. Citations in MLA format. Addresses a counterargument. Logical organization with transitions. Conclusion tied to policy impact.

Coverage

67%

Overall similarity

68

Weakest rubric item

Citations in MLA format.

Rubric items without a close match

Citations in MLA format. (68, partial) Logical organization with transitions. (76, partial)

The model

The files are served from this site, versioned so a cached copy is never stale. Nothing is fetched from a third party while you use the tool.

You can switch models in the tool. Also available: Gemma 4 E2B instruction-tuned (web build) (2.0 GB). The larger language model gives more thorough answers but is a bigger download and needs a stronger device.

Limitations

• A close match in meaning is not proof. A resume line that sounds like a requirement still has to be true, and a grounded-sounding sentence still has to be supported by the source.
• Very short items (one or two words) carry little meaning, so their scores are less reliable than full sentences.
• Results describe the input you gave, not the real-world situation behind it. Review them yourself before acting.

Good uses

• Triage education and study notes before deeper review
• Prepare consistent labels for repeat workflows
• Reduce manual sorting on small batches
• Keep sensitive drafts in the browser preview