Deep Learning Models for Automatic Summarization – Towards Data Science

https://towardsdatascience.com/deep-learning-models-for-automatic-summarization-4c2b89f2a9ea

Deep Learning Models for Automatic Summarization – Towards Data Science
Figure 1: Basic Seq2Seq encoder-decoder architecture with attention.The x_i are the input token embeddings, the a_i^t are the attention weights at step t, the h_i are the context vectors, h^t is the sentence embedding at step t obtained by weighting the context vectors with the attention weights, s_i are the decoder states, x’_i are the embeddings of the generated token (at inference time …
towardsdatascience.com

Publié

dans

par

Étiquettes :