Research map / Large language models / Language, safety and interpretability
Co-Attention Hierarchical Network: Generating Coherent Long Distractors for Reading Comprehension
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 article-question co-attention to HRED (ablation) | HRED distractor generation · RACE test, 1st Distractor | BLEU-4: 6.34 -> 6.85 | Helped |
| add semantic similarity loss to HRED (ablation) | HRED distractor generation · RACE test, 1st Distractor | BLEU-4: 6.34 -> 6.88 | Helped |
| remove the merging layer and use question-aware representation G as sentence representation (ablation) | Co-Attention Hierarchical Network · RACE test, 1st Distractor | BLEU-4: 6.85 -> 6.37 | Hurt |
| combine co-attention, semantic similarity loss, and the merging layer (ablation) | HRED distractor generation · RACE test, 1st Distractor | BLEU-4: 6.34 -> 7.01 | Helped |
| add a pointer-generator copy mechanism | HRED distractor generation · RACE test, 1st Distractor | BLEU-4: 6.34 -> 6.68 | No clear effect |
| use the proposed co-attention hierarchical network rather than HSA | RACE long distractor generation · RACE test, 1st Distractor | BLEU-4: 6.43 -> 7.01 | Helped |
| use the proposed model rather than HRED | RACE long distractor generation · RACE test, first 100 samples | Coherence: 7.38 -> 7.32 | Hurt |
| use the proposed model rather than HSA | RACE long distractor generation · RACE test, first 100 samples | Distracting Ability: 4.28 -> 5.23 | Helped |
Closest papers by meaning
- Generating Distractors for Reading Comprehension Questions from Real Examinations (AAAI 2019 · similarity 0.92)
- Multi-Mention Learning for Reading Comprehension with Neural Cascades (ICLR 2018 · similarity 0.88)
- Hierarchical Attention Networks for Sentence Ordering (AAAI 2019 · similarity 0.87)
- Densely Connected Attention Propagation for Reading Comprehension (NeurIPS 2018 · similarity 0.87)
- Top-H Decoding: Adapting the Creativity and Coherence with Bounded Entropy in Text Generation (NeurIPS 2025 · similarity 0.83)
- Coarse-to-Fine Highlighting: Reducing Knowledge Hallucination in Large Language Models (ICML 2024 · 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.