Model profile
Qwen 2 Vl 72B Instruct
2024-09-17release date
2eval lineages
Evidence summary
Published evidence spans 2 evals and 3 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #98 of 258); its weakest is SafeArena (normalized_safety_score, #5 of 5).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗Qwen/Qwen2-VL-72B-Instruct
- 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 |
|---|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #148 / 260 | 14.73 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #184 / 260 | 84.67 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #163 / 260 | 58.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #98 / 258 | 97.59 | ↑ higher | Source ↗official | |
| SafeArenanormalized_safety_score | #5 / 5 | 21.5 | ↑ higher | Source ↗official |