Argument Revision on Human Evaluation Set 200 edit suggestions
4.41SimilarityGRPOfull
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GRPOfullContext=approach2026.04 | 4.41 | 4.46 | 4.24 | 4.46 | |
| LLaMA2026.04 | 4.29 | 4.38 | 4.21 | 4.42 | |
| PPOapp=sim2026.04 | 3.13 | 3.48 | 3.75 | 3.97 |
Loading the SOTA2 catalog…