CharRNN-pytorch

Pytorch implementation of Character level recurrent neural network to generate text

train.py

Use your own text data and name it data.txt Then run

python train.py

Your model will be saved in pretrained folder

Argument	Parse	Default
Embedding Dimension	--embed_dim	65
Dropout	--dropout	0.1
Hidden Size	--hidden_size	50
Number of LSTM layers	--num_layers	2
Batch Size	--batch_size	100
Chunk Size	--chunk_size	200
Epochs	--epochs	200
Learning Rate	--lr	0.01
Root directory of data	--root	data.txt
Save the model every N epochs	--save_every	50

generate.py

Arguement	Parse	Default
Initial Text	--initialtext	Where
Temperature	--temperature	1
Number of Letters to Generate	--n	100
Root directory of data	--root	data.txt

Output

python generate.py --n 200

Where of call, let all not no Risent us thy do.

BETHE:
Ay, that we onoth be miled to did; and
That here and manesty
Nabtalmy most evancion they stash?
You bick'd gelolot part? then bloinopt.        

LADY MACBETH