Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning to Write Stories with Thematic Consistency and Wording Novelty
Juntao Li, Lidong Bing, Lisong Qiu, Dongmin Chen, Dongyan Zhao, Rui Yan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add cache module (ablation) | CVAE · ROCStories | Human consistency: 1.47 -> 2.55 | Improved |
| add cache module (ablation) | AS2S · ROCStories | Human consistency: 1.13 -> 2.32 | Improved |
| add cache module (ablation) | CVAE · ROCStories | BLEU-1: 25.5 -> 27.8 | Improved |
| add cache module (ablation) | AS2S · ROCStories | BLEU-1: 19.7 -> 25.7 | Improved |
| add cache module (ablation) | CVAE · ROCStories | Distinct-4: 78.7 -> 81.6 | Mixed |
| add cache module (ablation) | CVAE · ROCStories | Human wording: 2.05 -> 1.61 | Hurt |
| add cache module (ablation) | CVAE · ROCStories | Human readability: 2.65 -> 2.68 | No clear effect |
| add CVAE latent variable instead of plain RNN (ablation) | AS2S-Cache · ROCStories | Distinct-4: 63.6 -> 81.6 | Improved |
| add CVAE (no cache) over C-LM / Cont-AS2S (ablation) | Cont-AS2S · ROCStories | Human consistency: 2.19 -> 1.47 | Hurt |
| cache size 20-30 versus 50-70 (ablation) | CVAE-Cache · ROCStories | BLEU-1: 27.8 -> 24.6 (size 20 vs 70) | Improved |
| CVAE-Cache versus Hierarchical | Hierarchical · WritingPrompts | Distinct-4: 40.2 -> 59.2 | Improved |
| CVAE-Cache versus Hierarchical | Hierarchical · WritingPrompts | BLEU-1: 31.1 -> 30.5 | No clear effect |
Closest papers by meaning
- One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt (ICLR 2025 · similarity 0.86)
- Generative Caching for Structurally Similar Prompts and Responses (NeurIPS 2025 · similarity 0.84)
- Agents' Room: Narrative Generation through Multi-step Collaboration (ICLR 2025 · similarity 0.84)
- Parameterized Synthetic Text Generation with SimpleStories (NeurIPS 2025 · similarity 0.84)
- Storybooth: Training-Free Multi-Subject Consistency for Improved Visual Storytelling (ICLR 2025 · similarity 0.83)
- Consistent Story Generation: Unlocking the Potential of Zigzag Sampling (NeurIPS 2025 · similarity 0.83)
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.
