Research map / Large language models / Retrieval and knowledge
Improving Attributed Long-form Question Answering with Intent Awareness
Xinran Zhao, Aakanksha Naik, Jay DeYoung, Joseph Chee Chang, Jena D. Hwang, Tongshuang Wu, Varsha Kishore
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 verbalized intents at inference | o3 · SQA-CS-V2 | Overall: 85.1 -> 86.0 | Improved |
| add verbalized intents at inference | gemini-2.5-pro · SQA-CS-V2 | Overall: 88.1 -> 89.7 | Improved |
| add verbalized intents at inference | claude-opus-4 · SQA-CS-V2 | Overall: 85.4 -> 89.0 | Improved |
| add verbalized intents at inference | o3 · DeepScholar Bench | Overall: 46.8 -> 43.2 | Hurt |
| add verbalized intents at inference | o3 · DeepScholar Bench | Citation precision: 39.1 -> 27.2 | Hurt |
| add verbalized intents at inference | gemini-2.5-pro · DeepScholar Bench | Overall: 54.8 -> 57.8 | Improved |
| use paragraph-only intents instead of all intents for o3 (ablation) | o3 · DeepScholar Bench | Overall: 43.2 -> 49.3 | Improved |
| use intent-multiview SFT instead of baseline SFT | Qwen3-8B · SQA-CS-V2 | Overall: 83.2 -> 88.6 | Improved |
| use intent-multiview SFT instead of baseline SFT | Llama3.1-8B · SQA-CS-V2 | Overall: 84.4 -> 89.2 | Improved |
| use explicit instead of implicit intent SFT (ablation) | Qwen3-4B · SQA-CS-V2 | Overall: 85.2 -> 87.5 | Improved |
| add paragraph intents to citation intents (ablation) | gemini-2.5-pro · SQA-CS-V2-dev | Overall: 88.6 -> 89.7 | Improved |
| add citation intents to paragraph intents (ablation) | gemini-2.5-pro · SQA-CS-V2-dev | Overall: 89.1 -> 89.7 | Improved |
Closest papers by meaning
- Chain-of-Thought Improves Text Generation with Citations in Large Language Models (AAAI 2024 · similarity 0.88)
- Precise Information Control in Long-Form Text Generation (NeurIPS 2025 · similarity 0.86)
- BABILong: Testing the Limits of LLMs with Long Context Reasoning-in-a-Haystack (NeurIPS 2024 · similarity 0.86)
- Long-form factuality in large language models (NeurIPS 2024 · similarity 0.86)
- Towards Verifiable Text Generation with Generative Agent (AAAI 2025 · similarity 0.86)
- Physics of Language Models: Part 3.1, Knowledge Storage and Extraction (ICML 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.
