Function Calling on BFCL (BFCL Metric)
46.9BFCL ScoreFinTool-Qwen3-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| FinTool-Qwen3-8BModel Variant=domain-tuned2026.03 | 46.9 | |
| Qwen3-14BModel Variant=base2026.03 | 46.8 | |
| FinTool-Qwen3-14BModel Variant=domain-tuned2026.03 | 44.3 | |
| Qwen3-8BModel Variant=base2026.03 | 43 | |
| MENTORModel=LLaMA-3.1-8B2025.10 | 30.46 | |
| Strict RLModel=LLaMA-3.1-8B2025.10 | 28.92 | |
| Outcome-based RLModel=LLaMA-3.1-8B2025.10 | 28.71 | |
| MENTORModel=LLaMA-3.2-3B2025.10 | 26.31 | |
| Outcome-based RLModel=LLaMA-3.2-3B2025.10 | 25.62 | |
| Strict RLModel=LLaMA-3.2-3B2025.10 | 25.43 | |
| SFTModel=LLaMA-3.1-8B2025.10 | 24.78 | |
| SFTModel=LLaMA-3.2-3B2025.10 | 23.94 | |
| VanillaModel=LLaMA-3.1-8B2025.10 | 23.42 | |
| VanillaModel=LLaMA-3.2-3B2025.10 | 22.18 |