A implementation of SeqGAN, MLE, PG-BLEU and Scheduled Sampling in PyTorch. We compare the methods on synthetic dataset, and real dataset consisting of Barack Obama speeches.
- PyTorch v1 Stable
- Python 3.6
- CUDA at least 8.0 (For GPU)
The idea is from paper [SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient] (https://arxiv.org/pdf/1609.05473.pdf)
The code is written in PyTorch with SeqGAN is based on the implementation from https://github.com/ZiJianZhao/SeqGAN-PyTorch
$ python main.py