Reading Comprehension on BoolQ (5-shot)
63.9AccuracyMoE 128e8a1s
Evaluation results
| Method | Links | |
|---|---|---|
| MoE 128e8a1sModel=MoE 128e8a1s, Training Steps=100k2026.05 | 63.9 | |
| MoE 128e8a4g1sModel=MoE 128e8a4g1s, Training Steps=100k2026.05 | 63.4 | |
| DenseModel=Dense, Training Steps=100k2026.05 | 62.6 |