This repository contains the official PyTorch implementation of the paper:
“Disentangled Image-Text Classification: Enhancing Visual Representations with MLLM-driven Knowledge Transfer”
Pretrained Checkpoint:
We provide the pretrained checkpoint of MD-MLLM on the N24News Dataset for reproducing the results reported in our paper.
N24News Dataset (Accuracy: 86.08%): Download Checkpoint. Food-101 Dataset (Accuracy: 95.02%): Download Checkpoint.
You can use this checkpoint for evaluation or fine-tuning on related tasks.
Code Availability:
Additional code and resources will be released soon. Stay tuned!