Language model output embedding vector

Part 1 (2019) Advanced (Part 1 v3)
4

i wonder if the weight-tying used in seq2seq as described in this post could be used in this situation. it could be possible to tell that marvelous is closer to wonderful than badger