Model profile
Vicuna 13B
LMSYSdeveloper
2023-03-30release date
#217 / 309overall rank
3eval lineages
Evidence summary
Vicuna 13B has an estimated overall rank of #217; its 90% source-sensitivity interval is #143–#250. Its behavior-only rank is #221; company governance moves the combined estimate to #217. Published evidence spans 3 evals and 3 of 7 behavior components. Its strongest relative result is Fake Alignment (FINE) (multiple_choice_safe_decision_rate, #5 of 14); its weakest is SafetyBench (UB, #18 of 21).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗lmsys/vicuna-13b-v1.3
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
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 | Source |
|---|---|---|---|---|
| Fake Alignment (FINE)multiple_choice_safe_decision_rate | #5 / 14 | ↑58.67 | Source ↗official | |
| Fake Alignment (FINE)open_ended_safe_response_rate | #7 / 14 | ↑96 | Source ↗official | |
| S-Evalbase_en_overall | #13 / 22 | ↑57.1 | Source ↗official | |
| SafetyBenchEM | #14 / 21 | ↑65.9 | Source ↗official | |
| SafetyBenchIA | #13 / 21 | ↑71.4 | Source ↗official | |
| SafetyBenchMH | #14 / 21 | ↑77.5 | Source ↗official | |
| SafetyBenchOFF | #8 / 21 | ↑68.4 | Source ↗official | |
| SafetyBenchPH | #14 / 21 | ↑65.3 | Source ↗official | |
| SafetyBenchPP | #12 / 21 | ↑75.4 | Source ↗official | |
| SafetyBenchUB | #18 / 21 | ↑53 | Source ↗official |