Research map / Large language models / Language, safety and interpretability
HInT: Hypergraph Infusion at the Structural Layers Improves Table Understanding
Wonjin Lee, Soomi Jeong, Kwang In Kim
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 hypergraph infusion at structural layers | Qwen2.5-3B (SFT) · WikiTQ | EM (%): 65.44 -> 67.39 | Improved |
| add hypergraph infusion at structural layers | Qwen2.5-3B (SFT) · HiTab | EM (%): 70.12 -> 71.75 | Improved |
| add hypergraph infusion at structural layers | Qwen2.5-3B (SFT) · TabFact | accuracy (%): 85.25 -> 86.71 | Improved |
| add hypergraph infusion at structural layers | Qwen2.5-3B (SFT) · HybridQA | EM (%): 39.92 -> 40.99 | Improved |
| add hypergraph infusion at structural layers | Qwen2.5-3B (SFT) · FeTaQA | BLEU (%): 26.11 -> 26.58 | Improved |
| add external (hyper)graph encoders (HeGTa, LLaSA) to a decoder-only LLM | Qwen2.5-3B (SFT) · WikiTQ | EM (%): 65.44 -> 62.52 | Hurt |
| add structural-head knowledge: remove top-40 structural heads vs random heads (ablation) | Qwen2.5-3B · WikiTQ | EM (%): 47.51 -> 12.73 (structural) vs 32.29 (random) | Improved |
| add three extra transformer layers (capacity-matched baseline) (ablation) | Qwen2.5 (SFT) · WikiTQ | EM (%): 65.44 -> 64.43 | Hurt |
| inject structure at identified structural layers instead of random layers (ablation) | HInT · WikiTQ | EM (%): 65.71 -> 67.39 | Improved |
| inject structure at identified structural layers instead of all layers (ablation) | HInT · WikiTQ | EM (%): 66.37 -> 67.39 | Improved |
| increase structural layers from 0 to 12 (ablation) | HInT · HiTab | EM (%): 70.10 -> 71.84 | Improved |
| increase structural layers from 12 to 20 (ablation) | HInT · HiTab | EM (%): 71.84 -> 71.21 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HeGTa — HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding (AAAI 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding (AAAI 2025 · similarity 0.89)
- NeedleInATable: Exploring Long-Context Capability of Large Language Models towards Long-Structured Tables (NeurIPS 2025 · similarity 0.89)
- Table as a Modality for Large Language Models (NeurIPS 2025 · similarity 0.88)
- Chain-of-Table: Evolving Tables in the Reasoning Chain for Table Understanding (ICLR 2024 · similarity 0.87)
- Multimodal Tabular Reasoning with Privileged Structured Information (NeurIPS 2025 · similarity 0.87)
- SPHINX: Structural Prediction using Hypergraph Inference Network (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.
