Output Diversity on INFINITY-EVAL
39.61Distinct ScoreDeepSeek-14B
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeek-14Bsampling_setting=S-best2026.01 | 39.61 | |
| DeepSeek-14Bsampling_setting=Mixed2026.01 | 35.33 | |
| DeepSeek-14Bsampling_setting=S-non-en avg2026.01 | 31.84 | |
| Qwen3-32Bsampling_setting=Mixed2026.01 | 31.47 | |
| Qwen3-32Bsampling_setting=S-best2026.01 | 28.66 | |
| Qwen3-8Bsampling_setting=Mixed2026.01 | 28.13 | |
| Qwen3-32Bsampling_setting=S-non-en avg2026.01 | 27.52 | |
| Qwen3-14Bsampling_setting=S-best2026.01 | 27.07 | |
| Qwen3-32Bsampling_setting=S-en2026.01 | 27 | |
| Qwen3-14Bsampling_setting=Mixed2026.01 | 26.73 | |
| DeepSeek-14Bsampling_setting=S-en2026.01 | 25.27 | |
| Qwen3-8Bsampling_setting=S-best2026.01 | 24.51 | |
| Qwen3-14Bsampling_setting=S-non-en avg2026.01 | 22.6 | |
| Qwen3-8Bsampling_setting=S-non-en avg2026.01 | 22.54 | |
| Qwen3-8Bsampling_setting=S-en2026.01 | 20.67 | |
| Qwen3-14Bsampling_setting=S-en2026.01 | 20.4 |