Response Generation on McDiv nuggets ~1K no hds
0.532Spearman rhoSentBERT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SentBERT2026.06 | 0.532 | 75.8 | |
| BERTsts2026.06 | 0.393 | 67.6 | |
| C x a_nBackbone=Qwen2.5-3B, Permutations=50, Granularity=per-byte2026.06 | 0.345 | 64.9 | |
| a_nBackbone=Qwen2.5-3B, Permutations=50, Granularity=per-byte2026.06 | 0.225 | 61.9 | |
| CBackbone=Qwen2.5-3B, Permutations=50, Granularity=per-byte2026.06 | 0.082 | 54.5 | |
| distinct-n2026.06 | -0.002 | 50.7 |