Question Answering on LSAT (OOD)
26.58AccuracyL1-Max
Evaluation results
| Method | Links | ||
|---|---|---|---|
| L1-MaxBase Model=DeepScaleR-1.5B2026.02 | 26.58 | 2,321 | |
| L1-ExactBase Model=DeepScaleR-1.5B2026.02 | 25.19 | 1,374 | |
| ThinkPrune-Iter2kBase Model=DeepScaleR-1.5B2026.02 | 24.82 | 3,244 | |
| GRPOBase Model=DeepScaleR-1.5B2026.02 | 24.6 | 3,426 | |
| LACONICBase Model=DeepScaleR-1.5B2026.02 | 24.53 | 1,923 |