Mathematical Reasoning on GSM8K (Accuracy, Delta, Error Correction)
97.19AccuracySST
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SSTcompute=staged compute, imax=4, decoding=greedy argmax2026.04 | 97.19 | 2.43 | 46.38 | |
| Matched fine-tuned baselinecompute=one forward pass per token, decoding=greedy argmax2026.04 | 94.77 | — | — |