Benchmarking Retrieval-Augmented Generation in Multi-Modal Contexts

Source code for our paper : Benchmarking Retrieval-Augmented Generation in Multi-Modal Contexts

Click the links below to view our paper and datasets.

If you find this work useful, please cite our paper and give us a shining star 🌟

@misc{liu2025benchmarkingretrievalaugmentedgenerationmultimodal,
      title={Benchmarking Retrieval-Augmented Generation in Multi-Modal Contexts}, 
      author={Zhenghao Liu and Xingsheng Zhu and Tianshuo Zhou and Xinyi Zhang and Xiaoyuan Yi and Yukun Yan and Yu Gu and Ge Yu and Maosong Sun},
      year={2025},
      eprint={2502.17297},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2502.17297}, 
}

🎃 Overview

The $M^2RAG$ benchmark evaluates Multi-modal Large Language Models (MLLMs) by using multi-modal retrieved documents to answer questions. It includes four tasks: image captioning, multi-modal QA, fact verification, and image reranking, assessing MLLMs’ ability to leverage knowledge from multi-modal contexts.

The Multi-Modal Retrieval Augmented Instruction Tuning (MM-RAIT) method further adapts MLLMs to multi-modal in-context learning, enhancing their effectiveness in utilizing knowledge from these retrieval documents.

🎃 Requirement

🌵 1. Install the following packages using Pip or Conda under this environment

Python==3.10
Pytorch
transformers==4.44.2 (4.46.1 for finetune qwen2-vl)
clip
faiss==1.9.0
tqdm
numpy
base64
diffusers
flash-attn
xformers
llamafactory
accelerate
nltk
rouge_score
sklearn

We provide the version file requirements.txt of all our used packages, if you have any problems configuring the environment, please refer to this document.

🌵 2. Prepare the pretrained models: MiniCPM-V 2.6, Qwen2-VL, and VISTA

MM-RAIT is built on MiniCPM-V 2.6 and Qwen2-VL model. VISTA is used for retrieval multi-modal documents.

🎃 Reproduce MM-RAIT

🌵 Download Code & Dataset

• First, use git clone to download this project:

git clone https://github.com/NEUIR/M2RAG
cd M2RAG

• Second, you can either directly download and use M2RAG, or follow the instructions in 'data/data_preprocess' to build it step by step. Please place the downloaded dataset in the data folder. (❗️Note: For the imgs.tsv, you need to download the data from this link and run 7z x imgs.7z.001).

data/
└──m2rag/
    ├──fact_verify/
    ├──image_cap/
    ├──image_rerank/
    ├──mmqa/
    ├──imgs.lineidx.new
    └──imgs.tsv
    

🌵 Inference for Zero-Shot setting

Once the dataset and vanilla models are ready, you can follow the instructions below to reproduce our zero-shot results.

• Step 1: Encode the queries from the test set and the multi-modal corpus for each task.

cd script
bash get_embed_test.sh

• Step 2: Retrieve the topN most relevant multi-modal documents for each query.

bash retrieval_test.sh

• Step 3: Use the retrieved documents for vanilla RAG inference.

bash inference_cpmv.sh or bash inference_qwen.sh

For Image Reranking task, please use:

bash compute_ppl_minicpmv.sh or bash compute_ppl_qwen2vl.sh

🌵 Train MM-RAIT using MM-RAIT

Using the MiniCPM-V 2.6 models as an example, I will show you how to reproduce the results in this paper. The same is true for the Qwen2-VL. Also, we provide fine-tuned checkpoints. You can skip this step and proceed directly to inference.

• First step: Prepare the training data.

bash get_embed_train.sh
bash retrieval_train.sh
cd ../data/
bash finetune/construct_finetune_data.sh

• Second step: Fine-tune the MiniCPM-V model using LoRA.

cd ../script
bash finetune_cpmv.sh

• Final step: Use the fine-tuned model for inference.

bash inference_cpmv.sh

For Image Reranking task, please use:

bash compute_ppl_minicpmv.sh

🎃 Evaluate Generation Effectiveness

Go to the src/evaluation folder and evaluate model performance as follow:

• For Image Captioning and Muit-modal QA tasks, please use:

python generation.py --reference_file path_to_reference_data --candidate_file path_to_generation_data

• For Multi-Modal Factify Verification task, please use:

python evaluation/classification.py --true_file path_to_reference_data --pred_file path_to_generation_data

• For Image Reranking task, please use:

python -m pytorch_fid path/to/reference_images path/to/rerank_images

🎃 Contact

If you have questions, suggestions, and bug reports, please email:

[email protected]     [email protected]