Tool Learning on StableToolBench I3-Inst.
76SoPRGPT-4 (DFSDT)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPT-4 (DFSDT)Backbone=GPT-4, Reasoning Strategy=DFSDT2025.01 | 76 | 63.9 | |
| GPT-3.5 (DFSDT)Backbone=GPT-3.5, Reasoning Strategy=DFSDT2025.01 | 69.9 | 67.2 | |
| GPT-4 (Parallel)Backbone=GPT-4, Reasoning Strategy=Parallel2025.01 | 69.7 | 70.5 | |
| Qwen2.5 (Parallel)Backbone=Qwen2.5-7B-Instruct, Reasoning Strategy=Parallel2025.01 | 68.3 | 57.6 | |
| DTA-LlamaBackbone=Llama-2-7B, Reasoning Strategy=Parallel, Training Protocol=Fine-tuned on DTA-Tool2025.01 | 67.5 | 59 | |
| GPT-3.5 (Parallel)Backbone=GPT-3.5, Reasoning Strategy=Parallel2025.01 | 56.6 | 50.8 | |
| ToolLLAMA (DFSDT)Backbone=Llama-2-7B, Reasoning Strategy=DFSDT2025.01 | 53.6 | 50.8 | |
| GPT-3.5 (ReAct)Backbone=GPT-3.5, Reasoning Strategy=ReAct2025.01 | 48.6 | — | |
| GPT-4 (ReAct)Backbone=GPT-4, Reasoning Strategy=ReAct2025.01 | 42.6 | 54.1 | |
| ToolLLaMA† (DFSDT)Backbone=Llama-2-7B, Reasoning Strategy=DFSDT, Training Protocol=DTA-Tool*2025.01 | 33.3 | 26.2 | |
| ToolLLAMA (ReAct)Backbone=Llama-2-7B, Reasoning Strategy=ReAct2025.01 | 29.8 | 41 | |
| LLMCompilerBackbone=Llama-2-7B, Reasoning Strategy=Parallel2025.01 | 27 | 36.5 | |
| ToolLLaMA† (ReAct)Backbone=Llama-2-7B, Reasoning Strategy=ReAct, Training Protocol=DTA-Tool*2025.01 | 20.5 | 24.6 |