# LantaoYu/SeqGAN

Implementation of Sequence Generative Adversarial Nets with Policy Gradient

Repository: https://github.com/LantaoYu/SeqGAN
Canonical: https://ross.abutalabs.com/products/seqgan
Language: Python
License Family: other
Last push: 2019-03-10T15:29:18+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3636, "days_push": 2733, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2092, forks 695 (observed 2026-08-28T04:06:12.558183+00:00)

## What it is
A Python/TensorFlow implementation of SeqGAN, the Sequence Generative Adversarial Nets with Policy Gradient model from the AAAI-17 paper. It trains a generator of discrete token sequences adversarially, using a discriminator as reward signal via Monte Carlo search and policy gradient.

## Use cases
- reproduce the SeqGAN synthetic data experiments with oracle evaluation
- train a GAN to generate sequences of discrete tokens
- apply policy gradient adversarial training to text generation
- study reinforcement learning-based GAN training for sequences
- benchmark negative log-likelihood of sequence generators

## When to choose
- you need a reference implementation of the SeqGAN paper for research or study
- you want to experiment with adversarial training of discrete sequence generators using policy gradient
- you are working with legacy TensorFlow r1.x and Python 2.7 environments

## When to avoid
- you need a maintained library or modern framework support (requires TensorFlow r1.0.1 and Python 2.7)
- you want production text generation - modern transformer LLMs are far more effective
- you need a permissive license - the repository has no license, so reuse rights are unclear

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, deep-learning
- domain: machine-learning, deep-learning
- platform: python
- tags: gan, seqgan, policy-gradient, reinforcement-learning, text-generation, tensorflow, research-code, aaai-17, natural-language-processing, linux, gpu

## Member repositories
- LantaoYu/SeqGAN (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:12.558183+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:55:29.650384+00:00, confidence not recorded.
  - readme: https://github.com/LantaoYu/SeqGAN (fetched 2026-08-28T04:06:12.558183+00:00, sha fa9b9f2e3240)
- Data as of 2026-08-30T08:39:29.467469+00:00.