Model profile
Internlm Chat 20B
InternLMdeveloper
2023-09-18release date
#67 / 267overall rank
4eval lineages
Evidence summary
Internlm Chat 20B has an estimated overall rank of #67; its 90% source-sensitivity interval is #20–#185. Its behavior-only rank is #63; company governance moves the combined estimate to #67. Published evidence spans 4 evals and 4 of 7 behavior components. Its strongest relative result is SALAD-Bench (base_information_safety_harms, #1 of 33); its weakest is SALAD-Bench (mcq_representation_toxicity, #29 of 33).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗internlm/internlm-chat-20b
- 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 |
|---|---|---|---|---|---|
| Fake Alignment (FINE)multiple_choice_safe_decision_rate | #3 / 14 | 69.33 | ↑ higher | Source ↗official | |
| Fake Alignment (FINE)open_ended_safe_response_rate | #7 / 14 | 96 | ↑ higher | Source ↗official | |
| FLAMESdata_protection | #1 / 13 | 63.16 | ↑ higher | Source ↗official | |
| FLAMESfairness | #1 / 13 | 52.61 | ↑ higher | Source ↗official | |
| FLAMESlegality | #2 / 13 | 71.74 | ↑ higher | Source ↗official | |
| FLAMESmorality | #1 / 13 | 54.23 | ↑ higher | Source ↗official | |
| FLAMESsafety | #3 / 13 | 51.05 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #7 / 33 | 29.74 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #9 / 33 | 23.45 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #7 / 33 | 29.36 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #8 / 33 | 27.96 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #7 / 33 | 31.79 | ↑ higher | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #7 / 33 | 27.27 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #2 / 33 | 98.89 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #1 / 33 | 99.8 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_malicious_use | #5 / 33 | 98.44 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #6 / 33 | 97 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #8 / 33 | 93.39 | ↑ higher | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #2 / 33 | 96.36 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #29 / 33 | 3.333 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #27 / 33 | 4.167 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #29 / 33 | 3.974 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #29 / 33 | 3.333 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #29 / 33 | 3.229 | ↑ higher | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #27 / 33 | 7.778 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyinstruction_attack | #9 / 31 | 68.97 | ↑ higher | Source ↗official | |
| SuperCLUE Safetyresponsible_ai | #11 / 31 | 65.45 | ↑ higher | Source ↗official | |
| SuperCLUE Safetytraditional_safety | #25 / 31 | 68.09 | ↑ higher | Source ↗official |