Code for Multi-attention Recurrent Network - AAAI 2018
MARN is written in pytorch with some code fragments (1. LSTM Hybrid and 2. Multi Attention Head) copied from CMU-Multimodal SDK https://github.com/A2Zadeh/CMU-MultimodalSDK/.
- The code is developed with Python 3.6 and Pytorch 1.2.0 (with GPU support) on Linux
- For reasons of my convenience data is required to be
data_dir = ../../data-- errors might pop-up when other directories are used. - The model can be executed by running MARN.py.
Please contact Sunny Verma at [email protected] for reporting bug fixes or suggestions.