Awesome

MMICT: Boosting Multi-Modal Fine-Tuning with In-Context Examples

Source code for TOMM 2024 paper "MMICT: Boosting Multi-Modal Fine-Tuning with In-Context Examples" [arXiv preprint].

Environment

The required environment is included in requirements.txt.

Dataset Preparation

We train and test our model on:

• MSVD

• MSR-VTT

• COCO

• VQAv2

How to run

To train the model:

bash run.sh

Acknowledgments

We thank the developers of LAVIS, BLIP-2, CLIP, for their public code release.