Tool Retrieval and Function Selection on Taskbench DL
58.9Function Selection AccuracyDynamic LR (DTDR-L)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Dynamic LR (DTDR-L)LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 58.9 | 0.85 | 0.64 | |
| Dynamic DR (DTDR-C)LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 28.2 | 0.54 | 0.29 | |
| LRLLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 23.6 | 0.55 | 0.44 | |
| QTS (Paramanayakam et al. 2025)LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 18.9 | 0.4 | 0.26 | |
| DRLLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 14.7 | 0.36 | 0.18 | |
| BM-25LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 13.5 | 0.31 | 0.15 | |
| QTS (Gao et al. 2025)LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 13.3 | 0.17 | 0.06 | |
| QTS (Vanilla)LLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 7.5 | 0.16 | 0.07 | |
| Random GuessLLM=Qwen 3 0.6B, ICL Method=hard masking2025.12 | 2.4 | 0.15 | 0.07 |