Commonsense Reasoning on WNG-L
91.88AccuracyIn-Squeeze
Evaluation results
| Method | Links | |
|---|---|---|
| In-Squeezerank=128 to 1, schedule=Min steps2026.02 | 91.88 | |
| Cont-Squeezerank=128 to 1, additional_steps=2002026.02 | 91.58 | |
| Cont-Squeezerank=128 to 1, additional_steps=7002026.02 | 91.41 | |
| In-Squeezerank=128 to 1, schedule=Standard2026.02 | 91.07 | |
| Direct Fine-tuningrank=1, additional_steps=2002026.02 | 90.94 | |
| Direct Fine-tuningrank=1, additional_steps=02026.02 | 90.73 | |
| Cont-Squeezerank=128 to 1, additional_steps=02026.02 | 90.69 | |
| Direct Fine-tuningrank=1, additional_steps=7002026.02 | 90.6 | |
| In-Squeezereduction=128 ... -> 1, strategy=Standard2026.02 | 82.44 | |
| Direct Fine-tuningrank=1, training_steps=+200 steps2026.02 | 82.02 | |
| Direct Fine-tuningrank=1, training_steps=+0 steps2026.02 | 81.97 | |
| Direct Fine-tuningrank=1, training_steps=+700 steps2026.02 | 81.8 | |
| Cont-Squeezereduction=128 -> 1, training_steps=+700 steps2026.02 | 81.38 | |
| In-Squeezereduction=128 ... -> 1, strategy=Min steps2026.02 | 81.29 | |
| Cont-Squeezereduction=128 -> 1, training_steps=+200 steps2026.02 | 81.16 | |
| Cont-Squeezereduction=128 -> 1, training_steps=+0 steps2026.02 | 37.67 | |
| Zero-shotmode=0-S2026.02 | 27 |