Research map / Image generation / GANs
Long Text Generation via Adversarial Training with Leaked Information
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use leaked discriminator features with a hierarchical MANAGER–WORKER generator instead of RankGAN | LeakGAN · Synthetic data, length 20 | NLL: 8.247 -> 7.038 | Helped |
| Use leaked discriminator features with a hierarchical MANAGER–WORKER generator instead of RankGAN | LeakGAN · Synthetic data, length 40 | NLL: 9.958 -> 7.191 | Helped |
| Use LeakGAN instead of the strongest baseline | LeakGAN · EMNLP2017 WMT News | BLEU-2: 0.8590 -> 0.956 | Helped |
| Use LeakGAN instead of the strongest baseline | LeakGAN · EMNLP2017 WMT News | BLEU-3: 0.6015 -> 0.819 | Helped |
| Use LeakGAN instead of the strongest baseline | LeakGAN · EMNLP2017 WMT News | BLEU-4: 0.4541 -> 0.627 | Helped |
| Use LeakGAN instead of the strongest baseline | LeakGAN · EMNLP2017 WMT News | BLEU-5: 0.463 -> 0.498 | Helped |
| Use LeakGAN instead of RankGAN | LeakGAN · COCO Image Captions | BLEU-2: 0.850 -> 0.950 | Helped |
| Use LeakGAN instead of RankGAN | LeakGAN · COCO Image Captions | BLEU-3: 0.672 -> 0.880 | Helped |
| Use LeakGAN instead of RankGAN | LeakGAN · COCO Image Captions | BLEU-4: 0.557 -> 0.778 | Helped |
| Use LeakGAN instead of RankGAN | LeakGAN · COCO Image Captions | BLEU-5: 0.544 -> 0.686 | Helped |
| Use LeakGAN instead of RankGAN | LeakGAN · Chinese Poems | BLEU-2: 0.812 -> 0.881 | Helped |
| Use LeakGAN instead of SeqGAN | LeakGAN · WMT News | Turing test score: 0.236 -> 0.554 | Helped |
Closest papers by meaning
- Branch-GAN: Improving Text Generation with (not so) Large Language Models (ICLR 2024 · similarity 0.92)
- Adversarial Feature Matching for Text Generation (ICML 2017 · similarity 0.91)
- MaskGAN: Better Text Generation via Filling in the _______ (ICLR 2018 · similarity 0.89)
- TextGAIL: Generative Adversarial Imitation Learning for Text Generation (AAAI 2021 · similarity 0.89)
- Adversarial Text Generation via Feature-Mover's Distance (NeurIPS 2018 · similarity 0.89)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.