LantaoYu/SeqGAN
A TensorFlow implementation of SeqGAN for generating sequences of discrete tokens using GANs with policy gradient reinforcement learning.
Not currently ranked — collecting fresh signals.
star history
This repository implements Sequence Generative Adversarial Nets with Policy Gradient, applying GANs to discrete token sequence generation. The generator uses policy gradient where the discriminator provides reward signals passed back via Monte Carlo search. The code includes a two-stage training process: supervised pre-training with maximum likelihood estimation, followed by adversarial training to improve the generator.
Frequently asked
- What is LantaoYu/SeqGAN?
- A TensorFlow implementation of SeqGAN for generating sequences of discrete tokens using GANs with policy gradient reinforcement learning.
- Is SeqGAN open source?
- Yes — LantaoYu/SeqGAN is an open-source project tracked on heatdrop.
- What language is SeqGAN written in?
- LantaoYu/SeqGAN is primarily written in Python.
- How popular is SeqGAN?
- LantaoYu/SeqGAN has 2.1k stars on GitHub.
- Where can I find SeqGAN?
- LantaoYu/SeqGAN is on GitHub at https://github.com/LantaoYu/SeqGAN.