Tool Selection on BFCL v2 (test)
67.2Recall@1L_PORTS
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| L_PORTSEncoder=BGE, Best LLM=LLAMA3-GROQ-8B-Tool-Use2026.07 | 67.2 | 73.23 | 78.1 | 67.2 | 74.6 | 73.1 | 5.31 | 5.63 | |
| L_replugEncoder=BGE, Best LLM=LLAMA3-GROQ-8B-Tool-Use2026.07 | 66.17 | 73.2 | 77.82 | 66.17 | 72.92 | 74.31 | 4.86 | 5.13 | |
| L_PORTSEncoder=RoBERTa, Best LLM=LLAMA3-GROQ-8B-Tool-Use2026.07 | 58.12 | 69.21 | 73.52 | 58.12 | 68.38 | 69.22 | 60.3 | 59.24 | |
| L_replugEncoder=RoBERTa, Best LLM=LLAMA3-GROQ-8B-Tool-Use2026.07 | 53.97 | 64.88 | 68.39 | 53.97 | 62.61 | 64.93 | 55.76 | 54.5 |