Research map / Large language models / Retrieval and knowledge
Modeling Uncertainty Trends for Timely Retrieval in Dynamic RAG
Bo Li, Tian Tian, Zhenghua Xu, Hao Cheng, Shikun Zhang, Wei Ye
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 |
|---|---|---|---|
| use second difference of entropy instead of first difference (ablation) | ETC (Llama2-7b) · Avg. score | average score: 0.341 -> 0.344 | Improved |
| use second difference of entropy instead of first difference (ablation) | ETC (Llama2-7b) · HotpotQA | EM: 0.275 -> 0.288 | Improved |
| use second difference instead of first difference (ablation) | ETC (Llama2-7b) · 2WikiMultihopQA | F1: 0.364 -> 0.360 | Hurt |
| add dynamic smoothing (ablation) | ETC (Llama2-7b) · Avg. score | average score: 0.334 -> 0.344 | Improved |
| add dynamic smoothing (vs fixed weight 0.9) (ablation) | ETC (Llama2-7b) · HotpotQA | EM: 0.261 -> 0.288 | Improved |
| add dynamic smoothing (ablation) | ETC (Llama2-7b) · 2WikiMultihopQA | EM: 0.269 -> 0.271 | No clear effect |
| add dynamic smoothing (ablation) | ETC · Avg. retrieval count | retrievals: 1.34 -> 1.29 | Improved |
| add dynamic smoothing (ablation) | ETC · 2WikiMultihopQA (100 annotated) | redundant retrieval ratio: 0.91 -> 0.79 | Improved |
| add dynamic smoothing (ablation) | ETC · 2WikiMultihopQA (100 annotated) | delayed retrieval ratio: 0.22 -> 0.22 | No clear effect |
| ETC trend-based trigger instead of DRAGIN (ablation) | ETC · 2WikiMultihopQA (100 annotated) | delayed retrieval ratio: 0.33 -> 0.22 | Improved |
| ETC vs DRAGIN on BioASQ (ablation) | ETC (Vicuna-13b) · BioASQ | accuracy: 0.527 -> 0.689 | Improved |
| ETC vs DRAGIN on StrategyQA (ablation) | ETC (Llama2-7b) · StrategyQA | accuracy: 0.641 -> 0.650 | No clear effect |
Closest papers by meaning
- Worse than Zero-shot? A Fact-Checking Dataset for Evaluating the Robustness of RAG Against Misleading Retrievals (NeurIPS 2025 · similarity 0.88)
- Long-Context LLMs Meet RAG: Overcoming Challenges for Long Inputs in RAG (ICLR 2025 · similarity 0.88)
- Inference Scaling for Long-Context Retrieval Augmented Generation (ICLR 2025 · similarity 0.88)
- Adapting to Non-Stationary Environments: Multi-Armed Bandit Enhanced Retrieval-Augmented Generation on Knowledge Graphs (AAAI 2025 · similarity 0.88)
- Retrieval is Not Enough: Enhancing RAG through Test-Time Critique and Optimization (NeurIPS 2025 · similarity 0.87)
- RAGGED: Towards Informed Design of Scalable and Stable RAG Systems (ICML 2025 · similarity 0.87)
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.
