Model profile
Yi 34B Chat
01.AIdeveloper
2023-11-23release date
#150 / 267overall rank
8eval lineages
Evidence summary
Yi 34B Chat has an estimated overall rank of #150; its 90% source-sensitivity interval is #58–#199. Its behavior-only rank is #148; company governance moves the combined estimate to #150. Published evidence spans 8 evals and 5 of 7 behavior components. Its strongest relative result is CMoralEval (familial_morality, #1 of 26); its weakest is S-Eval (base_en_overall, #20 of 22).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗01-ai/Yi-34B-Chat
- OpenRouter ↗01-ai/yi-34b-chat
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
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 |
|---|---|---|---|---|---|
| AIRBench 2024 Safety Scenariossafety_scenarios | #60 / 80 | 0.536 | ↑ higher | Source ↗official | |
| ChiSafetyBenchharmful_response_rate | #8 / 14 | 0.65 | ↓ lower | Source ↗official | |
| ChiSafetyBenchmcq_score | #10 / 12 | 68.54 | ↑ higher | Source ↗official | |
| CMoralEvalfamilial_morality | #1 / 26 | 0.71 | ↑ higher | Source ↗official | |
| CMoralEvalinternet_ethics | #1 / 26 | 0.69 | ↑ higher | Source ↗official | |
| CMoralEvalpersonal_morality | #1 / 26 | 0.66 | ↑ higher | Source ↗official | |
| CMoralEvalprofessional_ethics | #1 / 26 | 0.7 | ↑ higher | Source ↗official | |
| CMoralEvalsocial_morality | #1 / 26 | 0.71 | ↑ higher | Source ↗official | |
| CRiskEvaldeception_willingness | #3 / 17 | 18.68 | ↓ lower | Source ↗official | |
| CRiskEvaldesire_for_resource | #9 / 17 | 32.56 | ↓ lower | Source ↗official | |
| CRiskEvalharmful_goal | #10 / 17 | 41.77 | ↓ lower | Source ↗official | |
| CRiskEvalimprovement_intent | #11 / 17 | 51.23 | ↓ lower | Source ↗official | |
| CRiskEvalmalicious_coordination | #8 / 17 | 10.33 | ↓ lower | Source ↗official | |
| CRiskEvalself_preservation | #12 / 17 | 39.99 | ↓ lower | Source ↗official | |
| CRiskEvalsituational_awareness | #14 / 17 | 67.23 | ↓ lower | Source ↗official | |
| S-Evalbase_en_overall | #20 / 22 | 39.3 | ↑ higher | Source ↗official | |
| SafeDialBenchaggression | #3 / 18 | 7.26 | ↑ higher | Source ↗official | |
| SafeDialBenchethics | #2 / 18 | 7.68 | ↑ higher | Source ↗official | |
| SafeDialBenchfairness | #9 / 18 | 7.337 | ↑ higher | Source ↗official | |
| SafeDialBenchlegality | #1 / 18 | 8.117 | ↑ higher | Source ↗official | |
| SafeDialBenchmorality | #1 / 18 | 7.52 | ↑ higher | Source ↗official | |
| SafeDialBenchprivacy | #1 / 18 | 7.88 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #9 / 33 | 24.14 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #7 / 33 | 27.36 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #11 / 33 | 22.76 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #9 / 33 | 26.81 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #10 / 33 | 22.6 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #8 / 33 | 23.81 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #21 / 33 | 91.73 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #21 / 33 | 93.23 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_malicious_use | #21 / 33 | 89.36 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #26 / 33 | 87.74 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #26 / 33 | 81.07 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #17 / 33 | 89.19 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #21 / 33 | 25 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #18 / 33 | 31.67 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #18 / 33 | 27.76 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #20 / 33 | 26.43 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #20 / 33 | 26.98 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #18 / 33 | 31.67 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyinstruction_attack | #3 / 31 | 72.41 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyresponsible_ai | #7 / 31 | 69.09 | ↑ higher | Source ↗official | |
| SuperCLUE Safetytraditional_safety | #10 / 31 | 78.72 | ↑ higher | Source ↗official |