Model profile
Yi 6B Chat
Evidence summary
Safety. Yi 6B Chat has an estimated Safety rank of #233; its 90% source-sensitivity interval is #134–#267. Its behavior-only rank is #237; company governance moves the combined estimate to #233. Published Safety evidence spans 5 eval lineages and 5 of 7 components. Its strongest relative result is CMoralEval (familial_morality, #4 of 26); its weakest is CRiskEval (self_preservation, #17 of 17).
Freedom. Yi 6B Chat has an estimated Freedom rank of #173; its 90% source-sensitivity interval is #113–#412. Published Freedom evidence spans 3 eval lineages and 1 of 1 components. Its strongest relative result is SALAD-Bench (base_representation_toxicity, #7 of 33); its weakest is SafeDialBench (privacy, #15 of 18).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗01-ai/Yi-6B-Chat
- 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 |
|---|---|---|---|---|
| ChiSafetyBenchharmful_response_rate | #11 / 14 | ↓0.87 | Source ↗official | |
| ChiSafetyBenchmcq_score | #5 / 12 | ↑86.01 | Source ↗official | |
| CMoralEvalfamilial_morality | #4 / 26 | ↑0.52 | Source ↗official | |
| CMoralEvalinternet_ethics | #5 / 26 | ↑0.5 | Source ↗official | |
| CMoralEvalpersonal_morality | #4 / 26 | ↑0.5 | Source ↗official | |
| CMoralEvalprofessional_ethics | #5 / 26 | ↑0.5 | Source ↗official | |
| CMoralEvalsocial_morality | #4 / 26 | ↑0.51 | Source ↗official | |
| CRiskEvaldeception_willingness | #12 / 17 | ↓26.92 | Source ↗official | |
| CRiskEvaldesire_for_resource | #15 / 17 | ↓40.53 | Source ↗official | |
| CRiskEvalharmful_goal | #12 / 17 | ↓42.13 | Source ↗official | |
| CRiskEvalimprovement_intent | #12 / 17 | ↓52.06 | Source ↗official | |
| CRiskEvalmalicious_coordination | #14 / 17 | ↓23.85 | Source ↗official | |
| CRiskEvalself_preservation | #17 / 17 | ↓45.21 | Source ↗official | |
| CRiskEvalsituational_awareness | #5 / 17 | ↓61.29 | Source ↗official | |
| SafeDialBenchaggression | #8 / 18 | ↑7.127 | Source ↗official | |
| SafeDialBenchethics | #9 / 18 | ↑7.577 | Source ↗official | |
| SafeDialBenchfairness | #10 / 18 | ↑7.277 | Source ↗official | |
| SafeDialBenchlegality | #11 / 18 | ↑7.887 | Source ↗official | |
| SafeDialBenchmorality | #15 / 18 | ↑7.123 | Source ↗official | |
| SafeDialBenchprivacy | #4 / 18 | ↑7.67 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #10 / 33 | ↑23.71 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #8 / 33 | ↑27.04 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #9 / 33 | ↑23.82 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #10 / 33 | ↑24.84 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #12 / 33 | ↑20.24 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #10 / 33 | ↑23.38 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #25 / 33 | ↑86.78 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #20 / 33 | ↑93.77 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #26 / 33 | ↑82.67 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #27 / 33 | ↑86.12 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #27 / 33 | ↑78.37 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #20 / 33 | ↑86.72 | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #28 / 33 | ↑5.833 | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #28 / 33 | ↑3.889 | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #27 / 33 | ↑5.962 | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #28 / 33 | ↑5.238 | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #27 / 33 | ↑4.792 | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #28 / 33 | ↑6.667 | 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 |
|---|---|---|---|---|
| ChiSafetyBenchharmful_response_rate | #4 / 14 | ↑0.87 | Source ↗official | |
| ChiSafetyBenchrefusal_rr1 | #4 / 14 | ↓71.21 | Source ↗official | |
| ChiSafetyBenchrefusal_rr2 | #4 / 14 | ↓70.78 | Source ↗official | |
| SafeDialBenchaggression | #10 / 18 | ↓7.127 | Source ↗official | |
| SafeDialBenchethics | #10 / 18 | ↓7.577 | Source ↗official | |
| SafeDialBenchfairness | #9 / 18 | ↓7.277 | Source ↗official | |
| SafeDialBenchlegality | #8 / 18 | ↓7.887 | Source ↗official | |
| SafeDialBenchmorality | #4 / 18 | ↓7.123 | Source ↗official | |
| SafeDialBenchprivacy | #15 / 18 | ↓7.67 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #24 / 33 | ↓23.71 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #26 / 33 | ↓27.04 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #25 / 33 | ↓23.82 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #24 / 33 | ↓24.84 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #22 / 33 | ↓20.24 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #24 / 33 | ↓23.38 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #9 / 33 | ↓86.78 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #14 / 33 | ↓93.77 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #8 / 33 | ↓82.67 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #7 / 33 | ↓86.12 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #7 / 33 | ↓78.37 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #14 / 33 | ↓86.72 | Source ↗official |