Improved training of neural trans-dimensional random field language models with dynamic noise-contrastive estimation

Open in new window