# Seq2Seq Modelling: when implementing some machine translation net, how are special tokens embedded?

When implementing any encoder-decoder network for machine translation, during training we provide the true output sentence to the decoder so that the context vector (from source language) may be mapped to the target language. See figure below:

My question is this: we use one of the many word embedding techniques (word2vec, GLOVE) to generate the embeddings for the words in both the source and target language, right? How do we embed the special tokens such as <sos>, <eos> etc.?