Process with end-of-sentence symbol

[tungpv85]

Hi guys.
I'm running Nicolas code and I have some concerns about the end-of-sentence symbol, i.e. "$$$"
What I understand is Nicolas put "$$$" at the end of each sentence. So my questions are:

1. If I understand correctly, In Nicolas code, seem that the "$$$" is treated similar to normal word ? That mean, word2vec model will generate a vector for "$$$" like other words. Does the symbol important, i.e. can I remove it from each sentence without affecting the performance?
2. If the symbol important, do you think how can I add the symbol to the speech data where each "sentence" is a sequence of feature vectors ? Do I just define a fix arbitrary vector as "$$$" and then add to the end of each "sentence" ?

Thank you in advance