Self-correction on Llama-3.3-70B math n=30 (failure pool)
86.7Correction Rate
L_MEMORY (Source-Conditioned Role Relabeling)
Updated 1mo ago
Evaluation Results
| Method | Links | |
|---|---|---|
| 86.7 | ||
2026.06 | 6.7 | |
2026.06 | 3.3 | |
2026.06 | 0 | |
2026.06 | 0 |
L_MEMORY (Source-Conditioned Role Relabeling)
| Method | Links | |
|---|---|---|
| 86.7 | ||
2026.06 | 6.7 | |
2026.06 | 3.3 | |
2026.06 | 0 | |
2026.06 | 0 |