README.md

---
license: apache-2.0
datasets:
- dongsheng/DTA-Tool
base_model:
- meta-llama/Llama-2-13b
---

## Model Description

<!-- Provide a longer summary of what this model is. -->
DTA_llama2_7b is from the paper "[Divide-Then-Aggregate: An Efficient Tool Learning Method via Parallel Tool Invocation](https://arxiv.org/abs/2501.12432)". 
It is a large language model capable of invoking tools and can parallel invoke multiple tools within a single round. 
The tool format it used is similar to OpenAI's Function Call.

## Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
The related code can be found in our GitHub [repository](https://github.com/Zhudongsheng75/Divide-Then-Aggregate).

## Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

The training data comes from our specially constructed [DTA-Tool](https://huggingface.co/datasets/dongsheng/DTA-Toolhttps://github.com/OpenBMB/ToolBench), which is derived from [ToolBench](https://github.com/OpenBMB/ToolBench).

## Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

### Testing Data

<!-- This should link to a Dataset Card if possible. -->

We evaluated the performance of DTA-Llama on [StableToolBench](https://github.com/THUNLP-MT/StableToolBench).

### Results

![result](result.png)

## Citation

<!-- If there is a paper or blog post introducing the model, the APA
![image/png](https://cdn-uploads.huggingface.co/production/uploads/630da0fae57da204209411d3/ViBSn34pV-4LWJkIpUvSr.png) that should go in this section. -->
```bibtex
@misc{zhu2025dividethenaggregateefficienttoollearning,
      title={Divide-Then-Aggregate: An Efficient Tool Learning Method via Parallel Tool Invocation}, 
      author={Dongsheng Zhu and Weixian Shi and Zhengliang Shi and Zhaochun Ren and Shuaiqiang Wang and Lingyong Yan and Dawei Yin},
      year={2025},
      eprint={2501.12432},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2501.12432}, 
}
```