README.md

---
task_categories:
- image-text-to-text
- visual-question-answering
language:
- en
tags:
- visual-question-answering
- multimodal
- reinforcement-learning
- visual-reasoning
- spatial-reasoning
- transit-maps
---

# ReasonMap-Plus Dataset

The **ReasonMap-Plus** dataset is an extended dataset introduced in the paper [RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning](https://huggingface.co/papers/2510.02240). It addresses the challenge of sparse rewards in fine-grained visual reasoning for multimodal large language models (MLLMs), particularly in structured and information-rich settings like transit maps. ReasonMap-Plus achieves this by introducing dense reward signals through Visual Question Answering (VQA) tasks, which enables effective cold-start training of fine-grained visual understanding skills. This repository contains the `ReasonMap-Plus` data for evaluation and `ReasonMap-Train` for RewardMap training.

- **Paper:** [RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning](https://huggingface.co/papers/2510.02240)
- **Project Page:** https://fscdc.github.io/RewardMap
- **Code:** https://github.com/fscdc/RewardMap

## Sample Usage

To get started with the `ReasonMap-Plus` dataset, follow these steps to install dependencies, download the data, and prepare it for training.

### 1. Install dependencies

If you face any issues with the installation, please feel free to open an issue on the GitHub repository.

```bash
pip install -r requirements.txt
```

### 2. Download the dataset

You can download [ReasonMap-Plus](https://huggingface.co/datasets/FSCCS/ReasonMap-Plus) (for evaluation) and [ReasonMap-Train](https://huggingface.co/datasets/FSCCS/ReasonMap-Train) (for RewardMap training) from Hugging Face or by running the following command:

```bash
python utils/download_dataset.py
```

Then, put the data under the folder `data`.

### 3. Prepare data for Supervised Fine-Tuning (SFT)

If you plan to use tools like [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) for SFT training, first prepare the datasets by running the following command:

```bash
python utils/prepare_data_for_sft.py --dataset_dir path/to/your_data
```

### 4. Data Format Example

Your data will be transferred into a format similar to this for SFT:

```json
  {
    "conversations": [
      {
        "from": "human",
        "value": "<image> Please solve the multiple choice problem and put your answer (one of ABCD) in one \"\\boxed{}\". According to the subway map, how many intermediate stops are there between Danube Station and lbn Battuta Station (except for this two stops)? \
A) 8 \
B) 1 \
C) 25 \
D) 12 \
"
      },
      {
        "from": "gpt",
        "value": "B"
      }
    ],
    "images": [
      "./maps/united_arab_emirates/dubai.png"
    ]
  },
```

### 5. Training Example

You can train the `RewardMap` model using the provided scripts:

```bash
# RewardMap training
bash scripts/reward_map.sh
```

## Citation

If you find this paper useful in your research, please consider citing our paper:

```bibtex
@article{feng2025rewardmap,
  title={RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning},
  author={Feng, Sicheng and Tuo, Kaiwen and Wang, Song and Kong, Lingdong and Zhu, Jianke and Wang, Huan},
  journal={arXiv preprint arXiv:2510.02240},
  year={2025}
}
```