Note that codes are adapted from https://github.com/williamSYSU/TextGAN-PyTorch.
- PyTorch >= 1.1.0
- Python 3.6
- Numpy 1.14.5
- CUDA 7.5+ (For GPU)
- nltk 3.4
- tqdm 4.32.1
- KenLM (https://github.com/kpu/kenlm)
-
Download stable release and unzip: http://kheafield.com/code/kenlm.tar.gz
-
Need Boost >= 1.42.0 and bjam
- Ubuntu:
sudo apt-get install libboost-all-dev - Mac:
brew install boost; brew install bjam
- Ubuntu:
-
Run within kenlm directory:
mkdir -p build cd build cmake .. make -j 4 -
pip install https://github.com/kpu/kenlm/archive/master.zip -
For more information on KenLM see: https://github.com/kpu/kenlm and http://kheafield.com/code/kenlm/
-
To run SeqGAN
cd run python run_seqgan.py --train_ratio {train ratio} --trial {trial} --aug {augmentation type}
-
To run RelGAN
cd run python run_relgan.py --train_ratio {train ratio} --trial {trial} --aug {augmentation type}
- train ratio - float
the proportion of training examples used to train GAN model
default - 1.0 e.g.) 0.05, 1.0, ... - trial - int,
the index of trial (used to the name of exp directory)
e.g.) 1, 2, 3, ... - augmentation type - str,
the augmentation transform apply to update GAN model. (The combination of 'mask', 'rand', 'swap'. Use 'noaug' to train without augmentation)
e.g.) 'noaug', 'swap', 'mask', 'swap_mask', 'rand_mask_swap', ...
Note that 'rand_mask_swap' means that the augmentations 'rand', 'mask', and 'swap' apply to a text sequentially.