Synthetic Data Quality Labeler

How Synthetic Data Quality Labeler works

Press Load model once: your browser downloads Gemma 4 E2B instruction-tuned (web build) (2.0 GB) from this site and keeps it in its cache. Then add your text and press Classify locally. The model runs on this device using your graphics card, 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.

Gemma 4 E2B instruction-tuned (web build) is a small open language model from Google. The tool sends it the instruction below plus your text, asks for a JSON answer with fixed fields, and shows each field as a card. It runs at temperature 0, so the same input gives the same answer on the same device, though different graphics cards can word it slightly differently.

Inputs longer than 6,000 characters are cut to that length before they reach the model, and the result says so. The model only sees what you paste: it does not look anything up.

The instruction the model gets

You review a sample of synthetic (generated) data records for quality problems before they are used for training or testing. Look for impossible or out-of-range values, duplicated or templated rows, missing or malformed fields, inconsistent formats, values that contradict each other within a row, unrealistic patterns, and real-looking personal data that should not be there. Quote the records and values exactly as written; do not count rows, total values or compute statistics.

What it returns

• Quality label: Exactly one of 'Usable', 'Usable after fixes' or 'Not usable', followed by a short reason.
• Problems found: Each problem written as 'issue type: the record or value that shows it'.
• What looks realistic: Parts of the data that look plausible and well formed.
• Fixes to the generator: Concrete changes to the generator or the data that would fix the problems.

Worked example

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

name,age,state,signup_date,plan John Smith,34,CA,2026-02-11,pro John Smith,34,CA,2026-02-11,pro Maria Lopez,152,TX,2026-02-30,basic Alex Kim,28,,2026-03-04,pro Test User,0,ZZ,2026-03-04,enterprise

Quality label

Not usable

Problems found

issue type: age 152 issue type: missing state issue type: missing name issue type: age 0 issue type: invalid state ZZ

What looks realistic

Valid date format in signup_date for some records Consistent structure of fields

Fixes to the generator

Fix age 152 to a realistic value. Populate the missing state for Alex Kim. Populate the missing name for Test User. Fix age 0 to a realistic value. Replace invalid state ZZ with a valid state abbreviation.

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.

Limitations

• Small language models make mistakes: they can miss an item, merge two, or state something more confidently than the text supports. Check the result against your input before you rely on it.
• It needs WebGPU and about 8 GB of memory, so it is offered on desktops and laptops, not on phones and tablets. Devices that cannot run it get the keyword fallback, clearly labelled.
• Results describe the input you gave, not the real-world situation behind it. Review them yourself before acting.

Good uses

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