Model profile
Evidence summary
Safety. Tulu 2 DPO 70B 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, #3 of 33); its weakest is SALAD-Bench (attack_enhanced_representation_toxicity, #24 of 33).
Freedom. Tulu 2 DPO 70B has an estimated Freedom rank of #321; its 90% source-sensitivity interval is #140–#571. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is SALAD-Bench (attack_enhanced_representation_toxicity, #10 of 33); its weakest is SALAD-Bench (base_representation_toxicity, #17 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-70b
- 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 | #24 / 33 | ↑8.19 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #19 / 33 | ↑8.14 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #24 / 33 | ↑8.16 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #24 / 33 | ↑8.22 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #24 / 33 | ↑7.35 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #20 / 33 | ↑10.82 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #18 / 33 | ↑94.35 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #17 / 33 | ↑94.65 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #19 / 33 | ↑92.75 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #20 / 33 | ↑92.96 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #17 / 33 | ↑88.87 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #21 / 33 | ↑86.49 | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #3 / 33 | ↑73.61 | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #4 / 33 | ↑64.72 | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #3 / 33 | ↑75.45 | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #3 / 33 | ↑68.1 | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #3 / 33 | ↑68.96 | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #3 / 33 | ↑66.67 | 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 | #10 / 33 | ↓8.19 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #15 / 33 | ↓8.14 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #10 / 33 | ↓8.16 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #10 / 33 | ↓8.22 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #10 / 33 | ↓7.35 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #14 / 33 | ↓10.82 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #15 / 33 | ↓94.35 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #17 / 33 | ↓94.65 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #15 / 33 | ↓92.75 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #14 / 33 | ↓92.96 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #17 / 33 | ↓88.87 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #13 / 33 | ↓86.49 | Source ↗official |