:question: Attention-based Visual Question Answering in Torch - View it on GitHub