1 paper
Junde Xu, Yuansheng Huang, Zijun Gao +5
Understanding and generation are often treated as two separate paradigms in training deep neural networks, despite the fact that both are trained with closely related objectives su…