Tsvetomila Mihaylova

latent structure model (2) transformer architecture (1) reinforcement learning (1) structured prediction (1) natural language processing (1) machine translation (1) natural language inference (1) semantic parsing (1) question classification (1) neural architecture (1) gradient approximation (1) surrogate gradient (1) discrete latent variable (1) straight-through estimator (1) neural network architecture (1) self-attention network (1) structured datum (1) exposure bia (1) sequence-to-sequence model (1) scheduled sampling (1)