Developer
LMSYS
8 indexed models; 5 currently meet the evidence threshold for the overall ranking. Together they have results from 9 evaluations.
Models by LMSYS
| Model | Evals | Components | Rank | Release date |
|---|---|---|---|---|
| Vicuna 13B | 3 | 3/7 | 189 | 2023-03-30 |
| Vicuna 7B V1.5 | 4 | 4/7 | 208 | 2023-06-22 |
| Vicuna 13B V1.5 | 4 | 4/7 | 222 | 2023-03-30 |
| Vicuna 7B | 4 | 3/7 | 236 | 2023-06-22 |
| Vicuna 33B V1.3 | 4 | 4/7 | 245 | 2023-06-22 |
| Vicuna 13B V1.5 16K | 1 | 4/7 | — | 2023-03-30 |
| Vicuna 33B | 1 | 3/7 | — | 2023-06-22 |
| Vicuna 7B V1.5 16K | 1 | 4/7 | — | 2023-06-22 |
Evaluations covering LMSYS models (9)
Do-Not-Answer · Fake Alignment (FINE) · HarmBench · JailBench · Large-scale Moral Machine experiment on LLMs · S-Eval · SafetyBench · SALAD-Bench · SORRY-Bench