self.normalizer = nn.Sigmoid() # D/TODO: Embed all the words and create the embedding matrix embs = self.word_emb.forwards(sent) # D/TODO: The second dimension should match...? assert embs.shape[1] == self.word_emb.embedding_size() # D/TODO: Calculate the matrix with the scores scores = self.normalizer(self.linear_layer(embs)) # D/TODO: The second dimension should match...? assert scores.shape[1] == len(self.tagset) # D/TODO: determine the position with the highest score _, ix = torch.max(score_vect, 0) ix = ix.item() # D/TODO: assert the index is within the range of POS tag indices assert 0 <= ix < len(self.tagset) # D/TODO: determine the corresponding POS tag pos = list(self.tagset)[ix]