Task Completion Rate on ToolBench
37.8Task Completion Rate (TCR)DPO – ADARUBRIC-DA
Evaluation Results
| Method | Links | |
|---|---|---|
| DPO – ADARUBRIC-DABackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 37.8 | |
| DPO – ADARUBRIC-GMBackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 35.6 | |
| DPO – ADARUBRIC-WMBackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 34.2 | |
| DPO – PrometheusBackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 29.3 | |
| DPO – G-EvalBackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 27.6 | |
| DPO – Random pairsBackbone=Qwen2.5-7B, Evaluation Protocol=DPO training2026.03 | 24.8 | |
| SFT – Success onlyBackbone=Qwen2.5-7B2026.03 | 23.1 | |
| BaseBackbone=Qwen2.5-7B, Evaluation Protocol=zero-shot2026.03 | 18.4 |