Research map / Large language models / Language, safety and interpretability
Order-Planning Neural Text Generation From Structured Data
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 |
|---|---|---|---|
| Add link-based attention to content-based attention using the adaptive hybrid gate (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 41.38 -> 43.01 | Helped |
| Add copy mechanism to content attention (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 41.38 -> 41.89 | Helped |
| Add copy mechanism to link attention (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 38.24 -> 39.08 | Helped |
| Use link attention alone instead of content attention (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 41.38 -> 38.24 | Hurt |
| Replace self-adaptive gating with fixed interpolation of content and link attention (ablation) | Order-planning model · WikiBio | BLEU-4: 42.89 BLEU when z=0.4 | Hurt |
| Feed weighted field information into content-attention computation (ablation) | Copy+Content attention model · WikiBio | BLEU-4: 41.89 -> 40.52 | Hurt |
| Feed weighted field information into the decoder RNN input (ablation) | Copy+Content attention model · WikiBio | BLEU-4: 41.89 -> 41.96 | Helped |
| Use proposed hybrid attention rather than feeding field information to the decoder input (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 41.96 -> 43.91 | Helped |
| Use the full order-planning model instead of content attention only (ablation) | WikiBio table-to-text generation · WikiBio | BLEU-4: 41.38 -> 43.91 | Helped |
| Use the full order-planning model instead of Table NLM | WikiBio table-to-text generation · WikiBio | BLEU-4: 34.70 -> 43.91 | Helped |
Closest papers by meaning
- Sentence Generation for Entity Description with Content-Plan Attention (AAAI 2020 · similarity 0.88)
- Data-to-Text Generation with Content Selection and Planning (AAAI 2019 · similarity 0.88)
- Order-Independence Without Fine Tuning (NeurIPS 2024 · similarity 0.85)
- Top-H Decoding: Adapting the Creativity and Coherence with Bounded Entropy in Text Generation (NeurIPS 2025 · similarity 0.85)
- Successor Features for Efficient Multi-Subject Controlled Text Generation (ICML 2024 · similarity 0.84)
- Decoding Game: On Minimax Optimality of Heuristic Text Generation Strategies (ICLR 2025 · similarity 0.84)
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.