Populate dataset card for RA-HMD with metadata, links, description, and usage (#1)

- Populate dataset card for RA-HMD with metadata, links, description, and usage (a5259321c47e47584f9dc234809600afdd3da499)


Co-authored-by: Niels Rogge <[email protected]>

README.md

@@ -0,0 +1,77 @@
+---
+license: cc-by-nc-4.0
+task_categories:
+  - image-text-to-text
+language:
+  - en
+tags:
+  - hateful-memes
+  - multimodal
+  - retrieval-augmented-generation
+  - vision-language
+---
+
+# RA-HMD Dataset
+
+This repository contains the dataset for the paper [Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection](https://huggingface.co/papers/2502.13061).
+
+This dataset supports the development of robust automated detection systems for hateful memes. It is designed to enhance in-domain accuracy and cross-domain generalization for Large Multimodal Models (LMMs) while preserving their general vision-language capabilities. The data provided includes the original datasets and a converted format suitable for use with the [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) framework for stage 1 training of the RA-HMD model.
+
+For more details and related resources:
+-   **Paper**: [Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection](https://huggingface.co/papers/2502.13061)
+-   **Code (GitHub)**: https://github.com/JingbiaoMei/RGCL
+-   **Project Page**: https://rgclmm.github.io/
+
+### Sample Usage
+
+The following instructions are derived from the [GitHub repository](https://github.com/JingbiaoMei/RGCL) and show how to set up the environment and generate embeddings for the dataset.
+
+#### Setup Environment for RA-HMD
+
+```bash
+git clone https://github.com/JingbiaoMei/RGCL.git
+cd RGCL/LLAMA-FACTORY
+conda create -n llamafact python=3.10
+conda activate llamafact
+pip install -e ".[torch,metrics,deepspeed,liger-kernel,bitsandbytes,qwen]"
+pip install torchmetrics wandb easydict
+pip install qwen_vl_utils torchvision
+# Install FAISS
+conda install -c pytorch -c nvidia faiss-gpu=1.7.4 mkl=2021 blas=1.0=mkl
+```
+
+#### Dataset Preparation - Generate CLIP Embedding
+
+First, ensure image data is copied into `./data/image/dataset_name/All` and annotation data (`jsonl`) into `./data/gt/dataset_name`. Then, generate CLIP embeddings:
+
+```shell
+python3 src/utils/generate_CLIP_embedding_HF.py --dataset "FB"
+python3 src/utils/generate_CLIP_embedding_HF.py --dataset "HarMeme"
+```
+
+#### Dataset Preparation - Generate ALIGN Embedding
+
+Similarly, generate ALIGN embeddings:
+
+```shell
+python3 src/utils/generate_ALIGN_embedding_HF.py --dataset "FB"
+python3 src/utils/generate_ALIGN_embedding_HF.py --dataset "HarMeme"
+```
+
+### Citation
+
+If you use this dataset in your research, please kindly cite the corresponding paper:
+
+```bibtex
+@article{RAHMD2025Mei,
+    title={Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection},
+    url={http://arxiv.org/abs/2502.13061},
+    DOI={10.48550/arXiv.2502.13061},
+    note={arXiv:2502.13061 [cs]},
+    number={arXiv:2502.13061},
+    publisher={arXiv},
+    author={Mei, Jingbiao and Chen, Jinghong and Yang, Guangyu and Lin, Weizhe and Byrne, Bill},
+    year={2025},
+    month=may
+}
+```