Research map / Large language models / Instruction and fine-tuning
Geometric Constraints for Small Language Models to Understand and Expand Scientific Taxonomies
Liri Fang, Dongqi Fu, Jiawei Han, Jingrui He, Vetle I Torvik
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 hyperbolic structure-dominated encoder (weight 0.5) (ablation) | SS-MONO (w/o AD) · SemEval-Food | MRR: 0.399 -> 0.430 | Improved |
| use hyperbolic encoder with cone loss instead of Euclidean (ablation) | SS-MONO · SemEval-Food | MRR: 0.175 -> 0.400 | Improved |
| add sequential self-attention in context encoder (ablation) | SS-MONO · SemEval-Food | MRR: 0.065 -> 0.430 | Improved |
| use self-attention instead of GAT (ablation) | SS-MONO · SemEval-Food | MRR: 0.215 -> 0.430 | Improved |
| use self-attention instead of GCN (ablation) | SS-MONO · SemEval-Food | MRR: 0.113 -> 0.430 | Improved |
| add structure loss on all neighbor types (1111) (ablation) | SS-MONO (w/o AD) · SemEval-Food | MRR: 0.217 -> 0.430 | Improved |
| sample 3 hops instead of 2 (ablation) | SS-MONO · SemEval-Food | MRR: 0.383 -> 0.430 | Improved |
| sample 3 hops instead of 4 (ablation) | SS-MONO · SemEval-Food | MRR: 0.406 -> 0.430 | Improved |
| add LLM augmented edge descriptions (ablation) | SS-MONO (w/o AD) · SemEval-Food | MR: 315.79 -> 239.17 | Improved |
| add LLM augmented descriptions (ablation) | SS-MONO (w/o AD) · WordNet-Verb | MRR: 0.297 -> 0.334 | Improved |
| add LLM augmented descriptions (ablation) | SS-MONO (w/o AD) · SemEval-Food | MRR: 0.430 -> 0.400 | Hurt |
| LLM reranking of top-100 candidates (ablation) | SS-MONO · SemEval-Food | MRR: 0.400 -> 0.462 | Improved |
Closest papers by meaning
- Structural Entropy Guided Agent for Detecting and Repairing Knowledge Deficiencies in LLMs (NeurIPS 2025 · similarity 0.87)
- End-to-End Ontology Learning with Large Language Models (NeurIPS 2024 · similarity 0.85)
- KG-FIT: Knowledge Graph Fine-Tuning Upon Open-World Knowledge (NeurIPS 2024 · similarity 0.85)
- SciLitLLM: How to Adapt LLMs for Scientific Literature Understanding (ICLR 2025 · similarity 0.85)
- Physics of Language Models: Part 3.1, Knowledge Storage and Extraction (ICML 2024 · similarity 0.84)
- SciInstruct: a Self-Reflective Instruction Annotated Dataset for Training Scientific Language Models (NeurIPS 2024 · similarity 0.84)
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.
