Program Repair on SpreadsheetBench (test)
84Pass RateIG
Evaluation Results
| Method | Links | |
|---|---|---|
| IGRound=A4, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 84 | |
| RLMRound=A3, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 84 | |
| IGRound=A2, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 81 | |
| RLMRound=A2, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 81 | |
| CC SubagentsRound=A3, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 81 | |
| IGRound=A3, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 80 | |
| IGRound=A5, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 80 | |
| Pure PatcherRound=A1, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 80 | |
| RLMRound=A1, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 79 | |
| IGRound=A1, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 77 | |
| CC SubagentsRound=A1, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 76 | |
| CC SubagentsRound=A2, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 71 | |
| Pure PatcherRound=A2, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 58 | |
| Pure PatcherRound=A3, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 58 | |
| CC SubagentsRound=A0, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 44 | |
| RLMRound=A0, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 42 | |
| IGRound=A0, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 41 | |
| Pure PatcherRound=A0, Patcher Model=Claude Code (Claude Opus 4.6)2026.05 | 41 |