No 01, Simplest classifier No 02, Pretrained Word2Vec initialization from Facebook's fasttext, kind of transfer learningish. Embedding layer is not trainable in this case and with trainable embedding ...