Research map / Large language models / Retrieval and knowledge
EchoEdit: Consistent Multi-Hop Question Answering via Ripple Control in Knowledge Editing
Jinwei Shi, Wenxuan Huang, Yu Xing, Yunhui Liu, Tao Zheng, Bin Chong, Tieke He
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 RippleGraph (ablation) | EchoEdit · MQuAKE-CF-3k-v2 / MQuAKE-T (Llama-3.1-8B) | accuracy: 32.8 -> 68.6 (CF); 41.9 -> 89.4 (T) | Improved |
| add RippleRule generator (ablation) | EchoEdit · MQuAKE-CF-3k-v2 / MQuAKE-T (Llama-3.1-8B) | accuracy: 51.3 -> 68.6 (CF); 74.1 -> 89.4 (T) | Improved |
| add CoT planner (ablation) | EchoEdit · MQuAKE-CF-3k-v2 / MQuAKE-T (Llama-3.1-8B) | accuracy: 56.5 -> 68.6 (CF); 78.9 -> 89.4 (T) | Improved |
| use EchoEdit instead of DecKER | Llama-3.1-8B-Instruct · MQuAKE-CF-3k-v2 | accuracy: 58.2 -> 68.6 | Improved |
| use EchoEdit instead of DecKER | Qwen2.5-7B-Instruct · RippleEdits-Random | accuracy: 31.5 -> 38.9 | Improved |
| use parameter editing (MEMIT) instead of graph/CoT-based methods | Llama-3.1-8B-Instruct · MQuAKE-CF-3k-v2 | accuracy: 7.1 vs 68.6 (EchoEdit) | Hurt |
| use EchoEdit on Qwen2.5-7B instead of Llama-3.1-8B | EchoEdit · MQuAKE-T | accuracy: 89.4 -> 74.8; EditCoT 75.3 on Qwen exceeds EchoEdit 74.8 | Mixed |
Closest papers by meaning
- Mitigating Error Accumulation in Knowledge Editing for Multi-Hop Question Answering (AAAI 2026 · similarity 0.92)
- Knowledge Editing with Dynamic Knowledge Graphs for Multi-Hop Question Answering (AAAI 2025 · similarity 0.90)
- Locate-then-edit for Multi-hop Factual Recall under Knowledge Editing (ICML 2025 · similarity 0.87)
- Uncovering Overfitting in Large Language Model Editing (ICLR 2025 · similarity 0.87)
- Tree-of-Reasoning Question Decomposition for Complex Question Answering with Large Language Models (AAAI 2024 · similarity 0.86)
- UniEdit: A Unified Knowledge Editing Benchmark for Large Language Models (NeurIPS 2025 · 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.
