Code Generation on MBPP 70% (test)
39.4AccuracyReflective Masking (Ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Reflective Masking (Ours)Base Model=LLaDA2026.06 | 39.4 | 8.8 | |
| Vanilla SFTBase Model=LLaDA2026.06 | 30.6 | — | |
| LLaDA2026.06 | 28 | — |
| Method | Links | ||
|---|---|---|---|
| Reflective Masking (Ours)Base Model=LLaDA2026.06 | 39.4 | 8.8 | |
| Vanilla SFTBase Model=LLaDA2026.06 | 30.6 | — | |
| LLaDA2026.06 | 28 | — |