# ColdGANs: Taming Language GANs with Cautious Sampling Strategies

arXiv: https://arxiv.org/abs/2006.04643 · HTML: https://mlsanity.com/paper/2006.04643/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 8 Jun 2020.

## Built on by (most cited first, 12 of 12)

- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Pretrained Language Models for Text Generation: A Survey](https://mlsanity.com/paper/2201.05273/) (2022)
- [Rethinking Automatic Evaluation in Sentence Simplification](https://mlsanity.com/paper/2104.07560/) (2021)
- [To Beam Or Not To Beam: That is a Question of Cooperation for Language GANs](https://mlsanity.com/paper/2106.06363/) (2021)
- [Reinforcement Learning for Generative AI: A Survey](https://mlsanity.com/paper/2308.14328/) (2023)
- [Reinforcement Learning for Generative AI: State of the Art, Opportunities and Open Research Challenges](https://mlsanity.com/paper/2308.00031/) (2023)
- [Generative Cooperative Networks for Natural Language Generation](https://mlsanity.com/paper/2201.12320/) (2022)
- [Improved Training of Mixture-of-Experts Language GANs](https://mlsanity.com/paper/2302.11875/) (2023)
- [Adaptive Bridge between Training and Inference for Dialogue](https://mlsanity.com/paper/2110.11560/) (2021)
- [Utilizing Local Hierarchy with Adversarial Training for Hierarchical Text Classification](https://mlsanity.com/paper/2402.18825/) (2024)
- [InitialGAN: A Language GAN with Completely Random Initialization](https://mlsanity.com/paper/2208.02531/) (2022)
- [OptAGAN: Entropy-based finetuning on text VAE-GAN](https://mlsanity.com/paper/2109.00239/) (2021)

## Abstract (arXiv)

Training regimes based on Maximum Likelihood Estimation (MLE) suffer from known limitations, often leading to poorly generated text sequences. At the root of these limitations is the mismatch between training and inference, i.e. the so-called exposure bias, exacerbated by considering only the reference texts as correct, while in practice several alternative formulations could be as good. Generative Adversarial Networks (GANs) can mitigate those limitations but the discrete nature of text has hindered their application to language generation: the approaches proposed so far, based on Reinforcement Learning, have been shown to underperform MLE. Departing from previous works, we analyze the exploration step in GANs applied to text generation, and show how classical sampling results in unstable training. We propose to consider alternative exploration strategies in a GAN framework that we name ColdGANs, where we force the sampling to be close to the distribution modes to get smoother learning dynamics. For the first time, to the best of our knowledge, the proposed language GANs compare favorably to MLE, and obtain improvements over the state-of-the-art on three generative tasks, namely unconditional text generation, question generation, and abstractive summarization.