Model profile
Baichuan 2 7B Chat
Baichuandeveloper
2023-09-06release date
#191 / 267overall rank
11eval lineages
Evidence summary
Baichuan 2 7B Chat has an estimated overall rank of #191; its 90% source-sensitivity interval is #119–#217. Its behavior-only rank is #194; company governance moves the combined estimate to #191. Published evidence spans 11 evals and 5 of 7 behavior components. Its strongest relative result is FLAMES (safety, #1 of 13); its weakest is CRiskEval (improvement_intent, #17 of 17).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗baichuan-inc/Baichuan2-7B-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 |
|---|---|---|---|---|---|
| ChineseSafescore | #13 / 22 | 53.99 | ↑ higher | Source ↗official | |
| ChiSafetyBenchharmful_response_rate | #8 / 14 | 0.65 | ↓ lower | Source ↗official | |
| ChiSafetyBenchmcq_score | #11 / 12 | 50.03 | ↑ higher | Source ↗official | |
| CMoralEvalfamilial_morality | #8 / 26 | 0.47 | ↑ higher | Source ↗official | |
| CMoralEvalinternet_ethics | #6 / 26 | 0.48 | ↑ higher | Source ↗official | |
| CMoralEvalpersonal_morality | #6 / 26 | 0.47 | ↑ higher | Source ↗official | |
| CMoralEvalprofessional_ethics | #6 / 26 | 0.49 | ↑ higher | Source ↗official | |
| CMoralEvalsocial_morality | #6 / 26 | 0.5 | ↑ higher | Source ↗official | |
| CRiskEvaldeception_willingness | #16 / 17 | 38.4 | ↓ lower | Source ↗official | |
| CRiskEvaldesire_for_resource | #14 / 17 | 39.06 | ↓ lower | Source ↗official | |
| CRiskEvalharmful_goal | #14 / 17 | 52.04 | ↓ lower | Source ↗official | |
| CRiskEvalimprovement_intent | #17 / 17 | 60.08 | ↓ lower | Source ↗official | |
| CRiskEvalmalicious_coordination | #16 / 17 | 29.29 | ↓ lower | Source ↗official | |
| CRiskEvalself_preservation | #13 / 17 | 40.2 | ↓ lower | Source ↗official | |
| CRiskEvalsituational_awareness | #6 / 17 | 61.78 | ↓ lower | Source ↗official | |
| DSPSafeBenchscore | #11 / 12 | 65.31 | ↑ higher | Source ↗official | |
| Fake Alignment (FINE)multiple_choice_safe_decision_rate | #12 / 14 | 20 | ↑ higher | Source ↗official | |
| Fake Alignment (FINE)open_ended_safe_response_rate | #5 / 14 | 97.33 | ↑ higher | Source ↗official | |
| FLAMESdata_protection | #8 / 13 | 40.79 | ↑ higher | Source ↗official | |
| FLAMESfairness | #4 / 13 | 42.17 | ↑ higher | Source ↗official | |
| FLAMESlegality | #4 / 13 | 52.17 | ↑ higher | Source ↗official | |
| FLAMESmorality | #11 / 13 | 39.3 | ↑ higher | Source ↗official | |
| FLAMESsafety | #1 / 13 | 56.41 | ↑ higher | Source ↗official | |
| HarmBenchdr | #13 / 28 | 18.8 | ↓ lower | Source ↗official | |
| M3-SafetyBenchoverall_score | #16 / 19 | 82.91 | ↑ higher | Source ↗official | |
| SafeDialBenchaggression | #11 / 18 | 7.073 | ↑ higher | Source ↗official | |
| SafeDialBenchethics | #6 / 18 | 7.613 | ↑ higher | Source ↗official | |
| SafeDialBenchfairness | #16 / 18 | 7.123 | ↑ higher | Source ↗official | |
| SafeDialBenchlegality | #9 / 18 | 7.937 | ↑ higher | Source ↗official | |
| SafeDialBenchmorality | #4 / 18 | 7.383 | ↑ higher | Source ↗official | |
| SafeDialBenchprivacy | #11 / 18 | 7.523 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyinstruction_attack | #15 / 31 | 65.52 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyresponsible_ai | #13 / 31 | 61.82 | ↑ higher | Source ↗official | |
| SuperCLUE Safetytraditional_safety | #5 / 31 | 80.85 | ↑ higher | Source ↗official |