Tool Use on BFCL Live Parallel
56.25BFCL Official Success RateQwen2.5-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-7BBackbone=Qwen2.5-7B-Base2026.06 | 56.25 | |
| SimpleRLBackbone=Qwen2.5-7B-Base2026.06 | 56.25 | |
| ReasonerBackbone=Qwen2.5-7B-Base2026.06 | 56.25 | |
| ISO-CBackbone=Qwen2.5-7B-Base2026.06 | 56.25 | |
| ISO-CTSBackbone=Qwen2.5-7B-Base2026.06 | 56.25 | |
| RESMERGEBackbone=Qwen2.5-7B-Base, configuration=SRC-A+LHC2026.06 | 56.25 | |
| TABackbone=Qwen2.5-7B-Base2026.06 | 50 | |
| TSV-MergeBackbone=Qwen2.5-7B-Base2026.06 | 50 | |
| ZeroBackbone=Qwen2.5-7B-Base2026.06 | 6.25 | |
| TIESBackbone=Qwen2.5-7B-Base2026.06 | 0 | |
| DARE + TIESBackbone=Qwen2.5-7B-Base2026.06 | 0 | |
| RAMBackbone=Qwen2.5-7B-Base2026.06 | 0 |