Text Revision on combined real-LLM corpus
10.74Mean Score (Comparator)Full-Restart
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Full-RestartMetric Category=waste2026.04 | 10.74 | — | 10.74 | — | 10 | 2.05 | 10.89 | |
| OracleMetric Category=quality2026.04 | 3.79 | 3.79 | — | — | 0.75 | 0.77 | 0.95 | |
| Full-RestartMetric Category=quality2026.04 | 3.16 | 3.16 | — | — | 0.11 | 0.12 | 0.27 | |
| NaiveMetric Category=quality2026.04 | 2.72 | 2.72 | — | — | 0.33 | 0.3 | 0.13 | |
| IgnoreMetric Category=quality2026.04 | 1.66 | 1.66 | — | — | 1.38 | 1.32 | 1.19 | |
| OracleMetric Category=waste2026.04 | 0 | — | 0 | — | 0.74 | 2.14 | 0.69 | |
| NaiveMetric Category=waste2026.04 | 0 | — | 0 | — | 0.74 | 2.38 | 0.69 | |
| AbsorberMetric Category=quality2026.04 | — | 3.04 | — | 3.04 | — | — | — | |
| AbsorberMetric Category=waste2026.04 | — | — | 0.74 | 0.74 | — | — | — |