Model profile
Llama 3.1 Nemotron Instruct 70B
2eval lineages
Evidence summary
Published evidence spans 2 evals and 4 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #14 of 260); its weakest is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #126 of 260).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗llama-3-1-nemotron-instruct-70b
- Official model catalog ↗Family-level model document · nvidia · first party
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 |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #95 / 311 | 0.6882 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #14 / 260 | 54.26 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #77 / 260 | 90.83 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #126 / 260 | 72.22 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #47 / 258 | 99.05 | ↑ higher | Source ↗official |