Long-context understanding on RULER (RULER Score)
91.42RULER ScoreDASH-3
Evaluation Results
| Method | Links | |
|---|---|---|
| DASH-3Budget=17.75, Search=12.3M / run, Train=700M2026.05 | 91.42 | |
| DASH-3Budget=9.0, Search=12.3M / run, Train=700M2026.05 | 87.63 | |
| Jet-Nemotron-2BSearch=200B, Train=400B2026.05 | 83.86 | |
| Jet-Nemotron-4BSearch=N/R, Train=400B2026.05 | 83.71 | |
| DASH-3Budget=4.875, Search=12.3M / run, Train=700M2026.05 | 73.58 |