Model profile
Evidence summary
Safety. Tulu 2 DPO 7B does not meet the evidence gate for a Safety rank. Published Safety evidence spans 1 eval lineages and 4 of 7 components. Its strongest relative result is SALAD-Bench (mcq_representation_toxicity, #18 of 33); its weakest is SALAD-Bench (attack_enhanced_representation_toxicity, #30 of 33).
Freedom. Tulu 2 DPO 7B has an estimated Freedom rank of #243; its 90% source-sensitivity interval is #78–#573. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is SALAD-Bench (attack_enhanced_representation_toxicity, #4 of 33); its weakest is SALAD-Bench (base_misinformation_harms, #11 of 33).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗allenai/tulu-2-dpo-7b
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
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 |
|---|---|---|---|---|
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #27 / 33 | ↑6.47 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #25 / 33 | ↑5.21 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #30 / 33 | ↑4.57 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #29 / 33 | ↑5.59 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #30 / 33 | ↑3.56 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #24 / 33 | ↑7.79 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #26 / 33 | ↑85.73 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #26 / 33 | ↑89.3 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #25 / 33 | ↑83.26 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #23 / 33 | ↑89.81 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #24 / 33 | ↑84.49 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #26 / 33 | ↑81.08 | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #19 / 33 | ↑27.78 | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #22 / 33 | ↑19.72 | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #20 / 33 | ↑26.86 | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #19 / 33 | ↑28.33 | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #18 / 33 | ↑28.65 | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #21 / 33 | ↑25.56 | 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 |
|---|---|---|---|---|
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #7 / 33 | ↓6.47 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #7 / 33 | ↓5.21 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #4 / 33 | ↓4.57 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #5 / 33 | ↓5.59 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #4 / 33 | ↓3.56 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #8 / 33 | ↓7.79 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #8 / 33 | ↓85.73 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #8 / 33 | ↓89.3 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #9 / 33 | ↓83.26 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #11 / 33 | ↓89.81 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #10 / 33 | ↓84.49 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #8 / 33 | ↓81.08 | Source ↗official |