Mathematical Reasoning on Countdown (CTD) (test)
43.8AccuracyNAR + Ours
Evaluation Results
| Method | Links | |
|---|---|---|
| NAR + OursSampling Regime=NAR, Time Schedule=+ Ours, Compute Budget=T = 322026.04 | 43.8 | |
| NARSampling Regime=NAR, Time Schedule=Linear, Compute Budget=T = 322026.04 | 42.2 | |
| NARSampling Regime=NAR, Time Schedule=Progressive, Compute Budget=T = 322026.04 | 39.5 | |
| SemiARSampling Regime=SemiAR, Time Schedule=Linear, Compute Budget=T = 322026.04 | 25.4 |