nn.Embedding using pre_trained weights question

I am trying to use the output of an LSTM net as the weight of a nn.Embedding layer. My question is, during the back propagation process, will my LSTM model gets optimized as well?