Tweeted By @adnothing
After an analysis of most DL frameworks followed by reimplementing (and painfully reproducing results of) YOLOv2 in both TensorFlow and @PyTorch in 2017, I made the decision to switch my team to PyTorch. We never looked back: UX and velocity is just 👍 #PyTorchFTW https://t.co/lpcaxBsSNQ
— Adrien Gaidon (@adnothing) January 26, 2019