1.
Neural seq2seq models are widely used to generate dialogue using variations of RNN Encoder-Decoder architecture with maximum likelihood estimation as objective function. [...]
2018 | masters theses |