1

I am confused on how LSTM learn from word embedding. I know that LSTM accepts 3D input (sample, timesteps, features). So, when we use embedding layer (word2vec) and we have 300-d vector representation, how does LSTM learn. I do understand that for a sentence, the words are timesteps. However, when we deal with vectors should the timesteps be same as vectors?

If it is true, then how can LSTM learn the sequence of words in that case to predict anything about sentence.

amy
  • 131
  • 3
  • @Sycorax It is not. I am trying to ask how the features are passed into LSTM unit and not the actual theory of LSTM units. Are LSTM units handling all the vectors/features together as neurons do? and then performing mathematical computations on these vectors? – amy Oct 06 '18 at 04:18
  • Have you read the article? It explains every step of the computation in LSTM units in detail. If you're not asking about how LSTMs do computation, please edit your question to explain what you are asking about. – Sycorax Oct 06 '18 at 14:21

0 Answers0