Hi, can you please let me know the model that you are using for training and also the multimodal loss function that you are using.