Where is this evidence ? Are you referring to this https://www.thelancet.com/journals/lancet/article/PIIS0140-6... ?
10 karma · joined February 29, 2016
Where is this evidence ? Are you referring to this https://www.thelancet.com/journals/lancet/article/PIIS0140-6... ?
The paper says
eij = a(si−1, hj) is an alignment model which scores how well the inputs around position j and the output at position i match. The score is based on the RNN hidden state si−1 (just before emitting yi) and the j-th annotation hj of the input sentence.
We parametrize the alignment model a as a feedforward neural network which is jointly trained with all the other components of the proposed system.
So a is a feed fw nn. What is the input to this nn ? Do we input each hj corresponding to each word of the input sentence separately and get one number for each word ? If the input sentence has 200 words do I run this 200 times ? One for each input word ?
The attention mechanism is a feed forward single layer neural network. The problem is that the input sequence is variable length. How are the outputs of the bi-rnn fed to the attention mechanism. What happens if I have a 80 word sentence as input and what happens if I have a 10 word sentence as input ?
Even if you don't know how to implement them you at least need to know how they work and their performance characteristics. I would not expect them to be able to implement one in C but at least they should what the fuck it is.
You are also obliged to pay for state health insurance in order to own a company which also changes constantly and upwards all the time. I you fail to pay your own health insurance you can go to jail. You cannot stop the health insurance. You have to close the company. If you owe health insurance payments and cannot pay them you cannot close the company and the health insurance debt keeps mounting up.