Model profile
Stanford Online All V4 S3
1eval lineages
Evidence summary
Published evidence spans 1 evals and 1 of 7 behavior components. Its strongest relative result is HELM Classic RealToxicityPrompts (toxic_prompt_toxic_fraction, #28 of 42); its weakest is HELM Classic RealToxicityPrompts (non_toxic_prompt_toxic_fraction, #35 of 42).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Model catalog entry ↗Exact model document · Stanford CRFM HELM · reliable secondary
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| HELM Classic RealToxicityPromptsnon_toxic_prompt_toxic_fraction | #35 / 42 | 0.02445 | ↓ lower | Source ↗official | |
| HELM Classic RealToxicityPromptstoxic_prompt_toxic_fraction | #28 / 42 | 0.09327 | ↓ lower | Source ↗official |