Posts

return state and return sequences

Image
 * Return Sequences:  it returns hidden state output of all time stamps * Return State: It returns three output. 1. hidden state of last time stamp 2. hidden state of last time stamp (again) 3. cell state of last time stamp. * Return State and Cell State (together) it returns three array. 1. in 1st array returns hidden state of all input time step. 2. return hidden state of last time step. 3. return cell state of last time step.  

What is word embedding?

Image
In word embedding, we convert sentences into vectors. This representation is dense and have semantic. word embedding is dense and semantic representation of words. here we can see than of x axis words have higher similarity score towards semantic y axis words. example. queen has higher score for royalty, human and living being than any other. similarly, man has higher score for gender and human.