Model profile
Chatglm2 Lite
2023-06-25release date
1eval lineages
Evidence summary
Published evidence spans 1 evals and 3 of 7 behavior components. Its strongest relative result is SafetyBench (PH, #2 of 21); its weakest is SafetyBench (UB, #15 of 21).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Official model documentation ↗Family-level model document · zai · first party
- 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 |
|---|---|---|---|---|---|
| SafetyBenchEM | #3 / 21 | 78.05 | ↑ higher | Source ↗official | |
| SafetyBenchIA | #2 / 21 | 85.05 | ↑ higher | Source ↗official | |
| SafetyBenchMH | #3 / 21 | 87.65 | ↑ higher | Source ↗official | |
| SafetyBenchOFF | #5 / 21 | 70.7 | ↑ higher | Source ↗official | |
| SafetyBenchPH | #2 / 21 | 79.65 | ↑ higher | Source ↗official | |
| SafetyBenchPP | #3 / 21 | 82.65 | ↑ higher | Source ↗official | |
| SafetyBenchUB | #15 / 21 | 59.15 | ↑ higher | Source ↗official |