Tool Learning on StableToolBench I2-Cat.
71.9SoPRDTA-Llama
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DTA-LlamaBackbone=Llama-2-7B, Reasoning Strategy=Parallel, Training Protocol=Fine-tuned on DTA-Tool2025.01 | 71.9 | 65.3 | |
| GPT-4 (Parallel)Backbone=GPT-4, Reasoning Strategy=Parallel2025.01 | 70.8 | 77.4 | |
| GPT-3.5 (DFSDT)Backbone=GPT-3.5, Reasoning Strategy=DFSDT2025.01 | 69.8 | 68.5 | |
| GPT-4 (DFSDT)Backbone=GPT-4, Reasoning Strategy=DFSDT2025.01 | 68 | 62.9 | |
| GPT-3.5 (Parallel)Backbone=GPT-3.5, Reasoning Strategy=Parallel2025.01 | 61.4 | 53.2 | |
| Qwen2.5 (Parallel)Backbone=Qwen2.5-7B-Instruct, Reasoning Strategy=Parallel2025.01 | 61.3 | 61.3 | |
| ToolLLAMA (DFSDT)Backbone=Llama-2-7B, Reasoning Strategy=DFSDT2025.01 | 53.4 | 49.2 | |
| GPT-4 (ReAct)Backbone=GPT-4, Reasoning Strategy=ReAct2025.01 | 48.9 | 62.1 | |
| GPT-3.5 (ReAct)Backbone=GPT-3.5, Reasoning Strategy=ReAct2025.01 | 43.9 | — | |
| ToolLLAMA (ReAct)Backbone=Llama-2-7B, Reasoning Strategy=ReAct2025.01 | 40.9 | 38.7 | |
| ToolLLaMA† (DFSDT)Backbone=Llama-2-7B, Reasoning Strategy=DFSDT, Training Protocol=DTA-Tool*2025.01 | 39.1 | 39.5 | |
| LLMCompilerBackbone=Llama-2-7B, Reasoning Strategy=Parallel2025.01 | 38.4 | 38.1 | |
| ToolLLaMA† (ReAct)Backbone=Llama-2-7B, Reasoning Strategy=ReAct, Training Protocol=DTA-Tool*2025.01 | 24.5 | 28.2 |