Function Calling / Tool Use on BFCL parallel parallel-multiple Actions (Accuracy)
82.2AccuracyBase
Evaluation Results
| Method | Links | |
|---|---|---|
| BaseModel=Qwen2.5-7B-Instruct, Evaluation Protocol=FULL2026.05 | 82.2 | |
| MAIGOModel=Qwen2.5-7B-Instruct, Evaluation Protocol=FULL2026.05 | 82.1 | |
| RLSTAModel=Qwen2.5-7B-Instruct, Evaluation Protocol=FULL2026.05 | 81.8 | |
| GRPOModel=Qwen2.5-7B-Instruct, Evaluation Protocol=FULL2026.05 | 81.6 | |
| SFTModel=Qwen2.5-7B-Instruct, Evaluation Protocol=FULL2026.05 | 79.3 | |
| MAIGOModel=Qwen2.5-3B-Instruct, Evaluation Protocol=FULL2026.05 | 77.5 | |
| RLSTAModel=Qwen2.5-3B-Instruct, Evaluation Protocol=FULL2026.05 | 73.2 | |
| BaseModel=Qwen2.5-3B-Instruct, Evaluation Protocol=FULL2026.05 | 72.8 | |
| GRPOModel=Qwen2.5-3B-Instruct, Evaluation Protocol=FULL2026.05 | 70.9 | |
| SFTModel=Qwen2.5-3B-Instruct, Evaluation Protocol=FULL2026.05 | 68.5 | |
| MAIGOModel=Qwen2.5-7B-Instruct, Evaluation Protocol=SHARDED2026.05 | 46.8 | |
| MAIGOModel=Qwen2.5-3B-Instruct, Evaluation Protocol=SHARDED2026.05 | 40 | |
| RLSTAModel=Qwen2.5-7B-Instruct, Evaluation Protocol=SHARDED2026.05 | 24.5 | |
| GRPOModel=Qwen2.5-7B-Instruct, Evaluation Protocol=SHARDED2026.05 | 21.5 | |
| SFTModel=Qwen2.5-7B-Instruct, Evaluation Protocol=SHARDED2026.05 | 18.2 | |
| BaseModel=Qwen2.5-7B-Instruct, Evaluation Protocol=SHARDED2026.05 | 17.8 | |
| RLSTAModel=Qwen2.5-3B-Instruct, Evaluation Protocol=SHARDED2026.05 | 17.3 | |
| GRPOModel=Qwen2.5-3B-Instruct, Evaluation Protocol=SHARDED2026.05 | 15.8 | |
| SFTModel=Qwen2.5-3B-Instruct, Evaluation Protocol=SHARDED2026.05 | 12.2 | |
| BaseModel=Qwen2.5-3B-Instruct, Evaluation Protocol=SHARDED2026.05 | 12 |