Model profile
Llama 3.1 Tulu 3 8B DPO
Evidence summary
Safety. Llama 3.1 Tulu 3 8B DPO does not meet the evidence gate for a Safety rank. Published Safety evidence spans 2 eval lineages and 5 of 7 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #38 of 246); its weakest is AbstentionBench (underspecified_context_f1, #18 of 20).
Freedom. Llama 3.1 Tulu 3 8B DPO has an estimated Freedom rank of #581; its 90% source-sensitivity interval is #364–#622. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #160 of 248); its weakest is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #208 of 246).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗allenai/Llama-3.1-Tulu-3-8B-DPO
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AbstentionBenchanswer_unknown_f1 | #10 / 20 | ↑0.8833 | Source ↗official | |
| AbstentionBenchfalse_premise_f1 | #12 / 20 | ↑0.668 | Source ↗official | |
| AbstentionBenchstale_f1 | #4 / 20 | ↑0.6833 | Source ↗official | |
| AbstentionBenchsubjective_f1 | #4 / 20 | ↑0.7733 | Source ↗official | |
| AbstentionBenchunderspecified_context_f1 | #18 / 20 | ↑0.4467 | Source ↗official | |
| AbstentionBenchunderspecified_intent_f1 | #8 / 20 | ↑0.7425 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #217 / 248 | ↑9.82 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #46 / 248 | ↑92.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #88 / 248 | ↑81.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #38 / 246 | ↑99.05 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #203 / 248 | ↓92.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #160 / 248 | ↓81.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #208 / 246 | ↓99.05 | Source ↗official |