LCS2-IIITD/ACL2024-MemeMQA

For the work on "MemeMQA: Multimodal Question Answering for Memes via Rationale-Based Inferencing"

ACL2024-MemeMQA

Paper

All models use hugging face checkpoints to get the pre-trained weights. Please note data paths are anonymised.

For implementation of the LLaVA model kindly refer to - https://github.com/haotian-liu/LLaVA

For implementation of Multimodal CoT model, kindly refer to - https://github.com/amazon-science/mm-cot