Mathematical Reasoning on MATH 500 (Accuracy, Delta, Error Correction)
89.8AccuracySST
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SSTcompute=staged compute, imax=4, decoding=greedy argmax2026.04 | 89.8 | 6.4 | 38.55 | |
| Matched fine-tuned baselinecompute=one forward pass per token, decoding=greedy argmax2026.04 | 83.4 | — | — |