Mathematical Reasoning on GSM8K (accuracy, average accuracy)
86.4AccuracyGDSD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GDSDSequence Length=2562026.05 | 86.4 | — | |
| GDSDSequence Length=5122026.05 | 86 | — | |
| GDSD w/ TLCSequence Length=2562026.05 | 85.6 | — | |
| DMPOSequence Length=5122026.05 | 85.2 | — | |
| SPGSequence Length=2562026.05 | 85.1 | — | |
| GDSD w/ TLCSequence Length=5122026.05 | 85 | — | |
| SPGSequence Length=5122026.05 | 84.9 | — | |
| GDSDSequence Length=1282026.05 | 83.9 | — | |
| ESPOSequence Length=5122026.05 | 83 | — | |
| SPGSequence Length=1282026.05 | 82.8 | — | |
| UniGRPOSequence Length=5122026.05 | 82.7 | — | |
| UniGRPOSequence Length=2562026.05 | 82.5 | — | |
| ESPOSequence Length=2562026.05 | 82.5 | — | |
| DMPOSequence Length=2562026.05 | 82.4 | — | |
| wd1Sequence Length=5122026.05 | 82.3 | — | |
| ESPOSequence Length=1282026.05 | 81.6 | — | |
| GDSD w/ TLCSequence Length=1282026.05 | 81.3 | — | |
| diffu-GRPO (d1)Sequence Length=5122026.05 | 81.2 | — | |
| wd1Sequence Length=2562026.05 | 80.8 | — | |
| LLaDA-8BSequence Length=5122026.05 | 80.2 | — | |
| diffu-GRPO (d1)Sequence Length=2562026.05 | 78.1 | — | |
| wd1Sequence Length=1282026.05 | 77.2 | — | |
| LLaDA-8BSequence Length=2562026.05 | 76.2 | — | |
| UniGRPOSequence Length=1282026.05 | 74.9 | — | |
| DMPOSequence Length=1282026.05 | 74.8 | — | |
| diffu-GRPO (d1)Sequence Length=1282026.05 | 74.6 | — | |
| LLaDA-8BSequence Length=1282026.05 | 71.3 | — | |
| diffu-GRPO (d1)Sequence Length=Average2026.05 | — | 78 | |
| DMPOSequence Length=Average2026.05 | — | 80.8 | |
| ESPOSequence Length=Average2026.05 | — | 82.4 | |
| GDSDSequence Length=Average2026.05 | — | 85.4 | |
| GDSD w/ TLCSequence Length=Average2026.05 | — | 84 | |
| LLaDA-8BSequence Length=Average2026.05 | — | 75.9 | |
| SPGSequence Length=Average2026.05 | — | 84.2 | |
| UniGRPOSequence Length=Average2026.05 | — | 80 | |
| wd1Sequence Length=Average2026.05 | — | 80.1 |