AI & Agents

Transformer Circuits

Last updated 2026-07-07 · benchmark checked 2026-08-09 · unchanged since 2026-08-08 — deterministic & reproducible

Research paper on verbalizable internal representations in language models.

Is Transformer Circuits production-ready?

Legit.Show scores Transformer Circuits 56 out of 100 — the simple average of its 7 measured frames. Legit.Show ran its deterministic 7-Frame production-readiness benchmark on Transformer Circuits (public-surface assessment), measured from the public surface with no LLM in the scoring path. Its strongest frame is Performance; its weakest is Privacy. Every frame it averages is published with its evidence on the Legit.Show listing.

The 7 Frames

• Performance — 96/100
• Accessibility — 67/100
• Security — 25/100
• Privacy — 0/100
• Reliability — 92/100
• Standards — 70/100
• Discoverability — 40/100

What we measured

• No Content-Security-Policy and no HSTS.
• Served over HTTPS with a valid certificate.
• Real Lighthouse performance run — 425 ms to first byte.
• Returns a proper 404 for unknown routes.
• 0 of 0 sampled routes reachable.
• No privacy policy found.
• Sets cookies / loads scripts with no consent prompt.
• Discoverable: OpenGraph image.

Who it's for

AI researchers · Machine learning practitioners · Interpretability researchers · Cognitive scientists · AI safety researchers

Visit Transformer Circuits → · Alternatives to Transformer Circuits → · How this was measured →