Unifying Event Detection and Captioning as Sequence Generation via Pre-Training [Paper] The code has been tested on PyTorch 1.7.1. Prerequisites Python 3.6 Pytorch >= 1.6.0 Usage For ANET, sh driver/run.sh