m-DAN: Dual Attention Networks for Multimodal Reasoning and Matching

Implementation of current SoTA model for image text matching task. Link to the paper: https://arxiv.org/abs/1611.00471

To train a model:

python main.py --mode 0

To test a model, make sure the model file is under models folder.

python main.py --mode 1 --model_file_name model_weights_1.t7