Research map / Large language models / Language, safety and interpretability
TabFact: A Large-scale Dataset for Table-based Fact Verification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 natural-language template linearization instead of cell concatenation (ablation) | Table-BERT, TabFact test · TabFact test | accuracy: 50.4 -> 65.1 | Helped |
| Use horizontal instead of vertical table scanning (ablation) | Table-BERT, TabFact test · TabFact test | accuracy: 56.2 -> 65.1 | Helped |
| Put table before fact rather than fact before table (ablation) | Table-BERT with horizontal template linearization · TabFact test | accuracy: 65.1 -> 65.1 | No clear effect |
| Add discriminator-based ranking of latent programs instead of equal-weight voting (ablation) | LPA, TabFact test · TabFact test | accuracy: 58.2 -> 65.0 | Helped |
| Use discriminator-weighted voting instead of equal-weight voting (ablation) | LPA, TabFact test · TabFact test | accuracy: 58.2 -> 63.1 | Helped |
| Feed the table caption to the LPA discriminator during ranking (ablation) | LPA-Ranking w/ Discriminator · TabFact test | accuracy: 65.0 -> 65.3 | No clear effect |
| Bootstrap NSM with LPA-guided programs before reinforcement learning (ablation) | NSM, TabFact test · TabFact test | accuracy: 54.1 -> 63.5 | Helped |
| Include table information rather than classifying the statement alone (ablation) | BERT classifier, TabFact test · TabFact test | accuracy: 50.5 -> 65.1 | Helped |
Closest papers by meaning
- Unsupervised Pretraining for Fact Verification by Language Model Distillation (ICLR 2024 · similarity 0.86)
- Heterogeneous Graph Reasoning for Fact Checking over Texts and Tables (AAAI 2024 · similarity 0.86)
- Multimodal Tabular Reasoning with Privileged Structured Information (NeurIPS 2025 · similarity 0.86)
- NeedleInATable: Exploring Long-Context Capability of Large Language Models towards Long-Structured Tables (NeurIPS 2025 · similarity 0.86)
- Table as a Modality for Large Language Models (NeurIPS 2025 · similarity 0.86)
- Chain-of-Table: Evolving Tables in the Reasoning Chain for Table Understanding (ICLR 2024 · similarity 0.85)
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.