---
license: apache-2.0
---

# Tiger Model Card

## Model details

Tactic-guided reasoner (Tiger) is a language model that solves *reasoning in the wild* task proposed in paper [Can LLMs Reason in the Wild with Programs](https://arxiv.org/abs/2406.13764).
It is trained by fine-tuning the LLaMA3-8B model on the [ReWild](https://huggingface.co/datasets/yuan-yang/ReWild) dataset.

**Model type:**
This repo contains the LoRA delta weights for `Tiger-PJ-8B`

We also provide the delta weights of other versions:
- [Tiger-Routing-8B](https://huggingface.co/yuan-yang/Tiger-Routing-8B/)
- [Tiger-PJ-8B](https://huggingface.co/yuan-yang/Tiger-PJ-8B)
- [Tiger-IPJ-8B](https://huggingface.co/yuan-yang/Tiger-IPJ-8B)

**License:**
Apache License 2.0

## Using the model

Check out how to use the model on our project page:  https://github.com/gblackout/Reason-in-the-Wild/


**Primary intended uses:**
Tiger is intended to be used for research.


## Citation

```
@article{yang2024can,
  title={Can LLMs Reason in the Wild with Programs?},
  author={Yang, Yuan and Xiong, Siheng and Payani, Ali and Shareghi, Ehsan and Fekri, Faramarz},
  journal={arXiv preprint arXiv:2406.13764},
  year={2024}
}
```