Model profile
GPT Neox 20B
2022-04-07release date
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, #17 of 42); its weakest is HELM Classic RealToxicityPrompts (non_toxic_prompt_toxic_fraction, #28 of 42).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗EleutherAI/gpt-neox-20b
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
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 | #28 / 42 | 0.02188 | ↓ lower | Source ↗official | |
| HELM Classic RealToxicityPromptstoxic_prompt_toxic_fraction | #17 / 42 | 0.07803 | ↓ lower | Source ↗official |