Update README.md

Update README.md

Train argument adjusted

Training and predicting succesfully

Fix typo, add comments

Bug fixxed, batchsize=100

Use less GPU resources

Set GRU to be bidirectional, Linear layer added

update comment

Predict successfully

Successfully trained

DataLoader: word_vec -> word_index with batch

modified data.py for seq2seq dataset

Create abstractive directory

Add argument [model saving directory]