Function Calling on BFCL (FN/SF Metrics)
0False Negative RateMPMA DPMA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MPMA DPMAModel=Qwen3-1.7B, Thinking Mode=Base2026.04 | 0 | 0 | |
| MPMA DPMAModel=Qwen3-8B, Thinking Mode=Think2026.04 | 0 | 0 | |
| MPMA DPMAModel=Qwen3-14B, Thinking Mode=Think2026.04 | 0 | 0 | |
| MPMA GAPMAModel=Qwen3-1.7B, Thinking Mode=Base2026.04 | 0 | 0 | |
| MPMA GAPMAModel=Qwen3-1.7B, Thinking Mode=Think2026.04 | 0 | 0 | |
| MPMA GAPMAModel=Qwen3-8B, Thinking Mode=Think2026.04 | 0 | 0 | |
| MPMA BenignModel=Llama-3B2026.04 | 1 | 1 | |
| MPMA BenignModel=Qwen3-1.7B, Thinking Mode=Base2026.04 | 1 | 1 | |
| MPMA BenignModel=Qwen3-1.7B, Thinking Mode=Think2026.04 | 1 | 1 | |
| MPMA BenignModel=Qwen3-8B, Thinking Mode=Think2026.04 | 1 | 1 | |
| MPMA BenignModel=Qwen3-14B, Thinking Mode=Think2026.04 | 1 | 1 | |
| MPMA DPMAModel=Llama-3B2026.04 | 1 | 1 | |
| MPMA DPMAModel=Qwen3-1.7B, Thinking Mode=Think2026.04 | 1 | 1 | |
| MPMA DPMAModel=Qwen3-14B, Thinking Mode=Base2026.04 | 1 | 1 | |
| MPMA GAPMAModel=Llama-3B2026.04 | 1 | 1 | |
| MPMA GAPMAModel=Qwen3-14B, Thinking Mode=Think2026.04 | 1 | 1 | |
| MPMA BenignModel=Qwen3-8B, Thinking Mode=Base2026.04 | 2 | 2 | |
| MPMA BenignModel=Qwen3-14B, Thinking Mode=Base2026.04 | 2 | 2 | |
| MPMA DPMAModel=Qwen3-8B, Thinking Mode=Base2026.04 | 2 | 2 | |
| MPMA GAPMAModel=Qwen3-8B, Thinking Mode=Base2026.04 | 2 | 2 | |
| MPMA GAPMAModel=Qwen3-14B, Thinking Mode=Base2026.04 | 2 | 2 | |
| MPMA BenignModel=Mistral-7B2026.04 | 7 | 7 | |
| MPMA GAPMAModel=Mistral-7B2026.04 | 9 | 9 | |
| MPMA DPMAModel=Mistral-7B2026.04 | 10 | 10 | |
| Function Injection (FS)Model=Mistral-7B2026.04 | 48 | 47 | |
| Function Injection (FS)Model=Qwen3-1.7B, Thinking Mode=Think2026.04 | 50 | 49 | |
| Function Injection (FS)Model=Llama-3B2026.04 | 57 | 57 | |
| Function Injection (FS)Model=Qwen3-1.7B, Thinking Mode=Base2026.04 | 64 | 64 | |
| Function Injection (FS)Model=Qwen3-8B, Thinking Mode=Base2026.04 | 64 | 64 | |
| Function Injection (FS)Model=Qwen3-8B, Thinking Mode=Think2026.04 | 65 | 64 | |
| Function Injection (FS)Model=Qwen3-14B, Thinking Mode=Base2026.04 | 68 | 67 | |
| Function Injection (FS)Model=Qwen3-14B, Thinking Mode=Think2026.04 | 68 | 66 | |
| FHAModel=Qwen3-8B, Thinking Mode=Base2026.04 | 70 | 65 | |
| FHAModel=Qwen3-8B, Thinking Mode=Think2026.04 | 70 | 65 | |
| Function Injection (ZS)Model=Mistral-7B2026.04 | 79 | 56 | |
| Function Injection (ZS)Model=Qwen3-14B, Thinking Mode=Think2026.04 | 79 | 47 | |
| Function Injection (ZS)Model=Llama-3B2026.04 | 80 | 59 | |
| Function Injection (ZS)Model=Qwen3-1.7B, Thinking Mode=Think2026.04 | 81 | 51 | |
| Function Injection (ZS)Model=Qwen3-8B, Thinking Mode=Think2026.04 | 81 | 47 | |
| FHAModel=Qwen3-14B, Thinking Mode=Base2026.04 | 82 | 73 | |
| FHAModel=Qwen3-14B, Thinking Mode=Think2026.04 | 82 | 73 | |
| Function Injection (ZS)Model=Qwen3-8B, Thinking Mode=Base2026.04 | 86 | 55 | |
| Standard InferenceModel=Llama-3B2026.04 | 88 | 88 | |
| Function Injection (ZS)Model=Qwen3-1.7B, Thinking Mode=Base2026.04 | 91 | 58 | |
| Function Injection (ZS)Model=Qwen3-14B, Thinking Mode=Base2026.04 | 91 | 57 | |
| Standard InferenceModel=Mistral-7B2026.04 | 96 | 96 | |
| Standard InferenceModel=Qwen3-1.7B, Thinking Mode=Base2026.04 | 96 | 96 | |
| Standard InferenceModel=Qwen3-1.7B, Thinking Mode=Think2026.04 | 97 | 97 | |
| Standard InferenceModel=Qwen3-8B, Thinking Mode=Base2026.04 | 97 | 97 | |
| FHAModel=Llama-3B2026.04 | 97 | 88 | |
| Standard InferenceModel=Qwen3-14B, Thinking Mode=Base2026.04 | 98 | 98 | |
| Standard InferenceModel=Qwen3-14B, Thinking Mode=Think2026.04 | 99 | 99 | |
| FHAModel=Mistral-7B2026.04 | 99 | 92 | |
| Standard InferenceModel=Qwen3-8B, Thinking Mode=Think2026.04 | 100 | 100 | |
| FHAModel=Qwen3-1.7B, Thinking Mode=Base2026.04 | 100 | 92 | |
| FHAModel=Qwen3-1.7B, Thinking Mode=Think2026.04 | 100 | 92 |