Question Answering on OOD Benchmarks Macro Average
32.26AccuracyGRPO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GRPOBase Model=DeepScaleR-1.5B2026.02 | 32.26 | 2,125 | |
| LACONICBase Model=DeepScaleR-1.5B2026.02 | 31.69 | 1,196 | |
| ThinkPrune-Iter2kBase Model=DeepScaleR-1.5B2026.02 | 31.42 | 1,882 | |
| L1-MaxBase Model=DeepScaleR-1.5B2026.02 | 31.26 | 1,652 | |
| L1-ExactBase Model=DeepScaleR-1.5B2026.02 | 28.25 | 1,205 |