Function Calling on BFCL Multi-Turn v4 (test)
46.75Overall AccFISSION-GRPO
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| FISSION-GRPOModel=Qwen3-8B2026.01 | 46.75 | 57.5 | 43.5 | 38 | 48 | |
| Dr.GRPOModel=Qwen3-8B2026.01 | 44.88 | 55 | 45 | 32.5 | 47 | |
| DAPOModel=Qwen3-8B2026.01 | 43.12 | 54.5 | 44.5 | 29 | 44.5 | |
| GRPOModel=Qwen3-8B2026.01 | 42.75 | 50.5 | 41 | 36 | 43.5 | |
| FISSION-GRPOModel=Qwen3-4B2026.01 | 40.87 | 51.5 | 42.5 | 30.5 | 39 | |
| DAPOModel=Qwen3-4B2026.01 | 38.25 | 48.5 | 36.5 | 28 | 40 | |
| BitAgent-8BModel=External 8B Agent Models2026.01 | 37.75 | 46.5 | 37.5 | 24 | 43 | |
| ToolACE-2-8BModel=External 8B Agent Models2026.01 | 37 | 47 | 31 | 28 | 42 | |
| GRPOModel=Qwen3-4B2026.01 | 36.38 | 46.5 | 34.5 | 27.5 | 37 | |
| Dr.GRPOModel=Qwen3-4B2026.01 | 34.75 | 43 | 34.5 | 27.5 | 34 | |
| BaseModel=Qwen3-8B2026.01 | 28.75 | 35.5 | 37 | 22.5 | 20 | |
| FISSION-GRPOModel=Qwen3-1.7B2026.01 | 20.38 | 29 | 18.5 | 16 | 18 | |
| BaseModel=Qwen3-4B2026.01 | 19.37 | 24.5 | 19 | 14.5 | 19.5 | |
| GRPOModel=Qwen3-1.7B2026.01 | 17.12 | 22 | 18.5 | 15.5 | 12.5 | |
| Dr.GRPOModel=Qwen3-1.7B2026.01 | 16.12 | 19.5 | 17.5 | 14.5 | 13 | |
| DAPOModel=Qwen3-1.7B2026.01 | 16 | 22 | 17 | 14 | 11 | |
| BaseModel=Qwen3-1.7B2026.01 | 7.8 | 10 | 11 | 8 | 2.5 |