Question Answering on MMLU OOD
43.36AccuracyLACONIC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LACONICBase Model=DeepScaleR-1.5B2026.02 | 43.36 | 497 | |
| GRPOBase Model=DeepScaleR-1.5B2026.02 | 43.3 | 719 | |
| ThinkPrune-Iter2kBase Model=DeepScaleR-1.5B2026.02 | 42.42 | 611 | |
| L1-MaxBase Model=DeepScaleR-1.5B2026.02 | 38.48 | 863 | |
| L1-ExactBase Model=DeepScaleR-1.5B2026.02 | 36.67 | 766 |