Research map / Large language models / Retrieval and knowledge
Following the Navigation: Enhancing Small Language Models Contextual Reasoning with LLM Guidance
Xiaoqi Ni, Jie Wang, Lin Yang, Yiyang Lu, Hanzhu Chen, Rui Liu, Jianye HAO
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 Navigation generation (LLM templates) (ablation) | Qwen2.5-3B-Instruct · HotpotQA | EM: 34.9 -> 51.1 | Improved |
| add Navigation generation (LLM templates) (ablation) | Qwen2.5-3B-Instruct · MuSR Team Allocation | accuracy: 34.5 -> 44.6 | Improved |
| add Navigation update (dynamic templates vs fixed general template) (ablation) | Qwen2.5-3B-Instruct · HotpotQA | EM: 35.1 -> 51.1 | Improved |
| add Navigation update (dynamic templates vs fixed general template) (ablation) | Qwen2.5-3B-Instruct · MuSR Object Placements | accuracy: 46.7 -> 52.6 | Improved |
| add Navigation generation (ablation) | Llama-3.2-3B-Instruct · StrategyQA | accuracy: 59.9 -> 68.7 | Improved |
| add Navigation update (ablation) | Llama-3.2-3B-Instruct · HotpotQA | EM: 36.1 -> 51.9 | Improved |
| add Navigation update (ablation) | Qwen2.5-7B-Instruct · HotpotQA | EM: 42.8 -> 52.3 | Improved |
| add Navigation templates (DS-R1) vs vanilla | Qwen2.5-3B-Instruct · HotpotQA | EM: 34.9 -> 51.1 | Improved |
| add Navigation templates vs CoT | Qwen2.5-7B-Instruct · HotpotQA | EM: 47.6 -> 52.3 | Improved |
| add Navigation to Llama-3.2-3B for cost study | Llama-3.2-3B-Instruct · MuSR | accuracy: 47.2 -> 54.3 | Improved |
| SLEICL few-shot rephrasing instead of vanilla | Qwen2.5-3B-Instruct · HotpotQA | EM: 34.9 -> 37.1 | Mixed |
| LoRA SFT instead of vanilla | Qwen2.5-7B-Instruct · MuSR Object Placements | accuracy: 47.3 -> 26.9 | Hurt |
Closest papers by meaning
- Supervised Knowledge Makes Large Language Models Better In-context Learners (ICLR 2024 · similarity 0.90)
- Large Language Models Think Too Fast To Explore Effectively (NeurIPS 2025 · similarity 0.89)
- Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes (ICLR 2026 · similarity 0.89)
- Information Re-Organization Improves Reasoning in Large Language Models (NeurIPS 2024 · similarity 0.89)
- ReGenesis: LLMs can Grow into Reasoning Generalists via Self-Improvement (ICLR 2025 · similarity 0.88)
- Mixture of In-Context Experts Enhance LLMs' Long Context Awareness (NeurIPS 2024 · similarity 0.88)
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.
