Analysis workspace

Target–auxiliary comparison

Black-box

Compares a target with a matched auxiliary model on identical prompts.

What this audit can surface: Is the suspected behavior stronger in the target than in a comparable clean or auxiliary model?
What would count as evidence: Differences are replayed on held-out prompts and interpreted only if the comparison is legitimate.
Execution requirement: The target plus a model matched closely enough for a defensible comparison.

Target against a defensible comparisonIdentical prompts, two models

TargetPrincipal-linked responses14 of 20

Illustrative held-out count

ComparisonPrincipal-linked responses1 of 20

Same prompts and settings

Illustrative target–comparison differential

References and technical details

Configure auditArtifact and tested scope

Hugging Face model ID

HFView on Hugging Face ↗

Audit configuration

Runs a versioned protocol with strict output parsing, matched controls, coverage gates, and uncertainty. Results are observations, not model-safety verdicts.

Audit promptsOne prompt per line. The result retains each prompt and response for review; it does not infer a hidden objective from text alone.Auxiliary or clean comparison modelIdentical prompts are sent to both models. Capability or lineage mismatch remains a limitation.

Up to 6 prompts × 2 models · Private

Uses Hugging Face Inference Providers under your account. Inference is billed to you.

Readiness

Sign in with Hugging Face to run this analysis.

Evidence boundary. Validity depends on the comparison model. Capability, lineage, or serving differences can masquerade as target-specific behavior.