Sudoku on Sudoku 4x4 (test)
14.7Accuracy (Avg)AR + beam-search
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| AR + beam-searchModel=LLaDA-1.52026.01 | 14.7 | 17.5 | 16.8 | 13.5 | 11 | — | — | |
| ARModel=LLaDA-1.52026.01 | 14.6 | 14.5 | 15.8 | 14.5 | 13.6 | — | — | |
| AR + MVModel=LLaDA-1.52026.01 | 14.4 | 14.6 | 16.6 | 13.2 | 13.1 | — | — | |
| AR + MVModel=LLaDA2026.01 | 14.2 | 14.4 | 15.9 | 12.1 | 14.3 | — | — | |
| ARModel=LLaDA2026.01 | 14.1 | 14.7 | 15.9 | 13.6 | 12 | — | — | |
| Low-conf + MVModel=LLaDA-1.52026.01 | 13 | 13.7 | 14.9 | 10.8 | 12.6 | — | — | |
| Order-Token SearchModel=LLaDA-1.52026.01 | 12.8 | 13.8 | 16.1 | 11.2 | 9.9 | — | — | |
| Low-confidenceModel=LLaDA-1.52026.01 | 12.7 | 13.5 | 15.6 | 11.4 | 10.3 | — | — | |
| AR + beam-searchModel=LLaDA2026.01 | 12.4 | 13.2 | 17.7 | 10.6 | 8.2 | — | — | |
| Random + MVModel=LLaDA-1.52026.01 | 10.2 | 11.1 | 11.3 | 9.3 | 9.2 | — | — | |
| OTS All BlocksModel=LLaDA-1.52026.01 | 9.8 | 9.3 | 10.5 | 9.6 | 9.7 | — | — | |
| Order-Token SearchModel=LLaDA2026.01 | 9.4 | 10.1 | 11.7 | 8.5 | 7.4 | — | — | |
| OTS Future BlocksModel=LLaDA-1.52026.01 | 9.4 | 9.6 | 11.1 | 8.3 | 8.5 | — | — | |
| Low-conf + MVModel=LLaDA2026.01 | 8.7 | 8.9 | 10.3 | 8.5 | 7.2 | — | — | |
| Low-confidenceModel=LLaDA2026.01 | 8.1 | 8.5 | 11.7 | 6.5 | 5.5 | — | — | |
| OTS Future BlocksModel=LLaDA2026.01 | 7.3 | 7 | 7.7 | 7.4 | 6.9 | — | — | |
| Random + MVModel=LLaDA2026.01 | 7.1 | 6.4 | 6.5 | 8.4 | 6.9 | — | — | |
| OTS All BlocksModel=LLaDA2026.01 | 6.9 | 6.5 | 6.9 | 7.5 | 6.9 | — | — | |
| CAPRBatch (s)=165, Update (s)=10.10, Conv. (h)=≈242026.06 | 0.93 | — | — | — | — | — | — | |
| d-TreeRPOBatch (s)=596, Update (s)=9.91, Conv. (h)=≈482026.06 | 0.929 | — | — | — | — | — | — | |
| GDPOBatch (s)=153, Update (s)=12.72, Conv. (h)=≈482026.06 | 0.257 | — | — | — | — | — | — | |
| TraceRLBatch (s)=604, Update (s)=43.14, Conv. (h)=≈482026.06 | 0.256 | — | — | — | — | — | — | |
| wd1Batch (s)=86, Update (s)=7.21, Conv. (h)=≈242026.06 | 0.252 | — | — | — | — | — | — | |
| SAPOBatch (s)=423, Update (s)=35.25, Conv. (h)=≈722026.06 | 0.203 | — | — | — | — | — | — | |
| Diffu-GRPOBatch (s)=111, Update (s)=9.12, Conv. (h)=≈242026.06 | 0.129 | — | — | — | — | — | — | |
| GPT-OSS-20BEvaluation protocol=Without task-specific program synthesis or symbolic search2026.03 | — | — | — | — | — | 100 | 100 | |
| HRMTraining Data=9x9 Sudoku, Inference Steps=16, Evaluation protocol=Zero-shot generalization2026.03 | — | — | — | — | — | 0 | 29.19 | |
| SE-RRMTraining Data=9x9 Sudoku, Inference Steps=16, Evaluation protocol=Zero-shot generalization2026.03 | — | — | — | — | — | 95.46 | 99.15 | |
| TRMTraining Data=9x9 Sudoku, Inference Steps=16, Evaluation protocol=Zero-shot generalization2026.03 | — | — | — | — | — | 0 | 45.88 |