Interactive tool-use on τ-bench retail domain
28.7Action Reward
GBC (Qwen-3-32B)
Updated 26d ago
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
2026.06 | 28.7 | 79.1 | 24.3 | |
2026.06 | 27.8 | 71.3 | 22.6 | |
2026.06 | 27 | 78.3 | 24.3 | |
2026.06 | 20 | 77.4 | 17.4 | |
2026.06 | 17.4 | 73 | 13.9 | |
2026.06 | 13.9 | 78.3 | 13 | |
2026.06 | 13.9 | 73 | 9.6 | |
2026.06 | 13.9 | 71.3 | 9.6 | |
2026.06 | 12.2 | 72.2 | 8.7 | |
2026.06 | 11.3 | 70.4 | 8.7 | |
2026.06 | 9.6 | 62.6 | 7 | |
2026.06 | 9.6 | 70.4 | 6.1 |