mlsanityData to 7 Oct 2026

Towards Robust Numerical Claim Verification

arXiv 2610.00689 ↗First posted 30 Sep 2026cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) are widely used for claim verification, yet remain brittle for numerical reasoning: even small changes in value can sharply degrade accuracy. We show that this brittleness persists in frontier LLMs, but can be mitigated through adversarial fine-tuning on numerically perturbed examples. Using parameter-efficient fine-tuning, small Qwen3 models (0.6Bunicodex20138B) reach 98.7% accuracy on label-flipping perturbations, outperforming larger zero-shot models and frontier systems (GPT-5.4 Pro (74.0%) and Gemini 2.5 Flash (73.9%)). The gains generalise to unseen perturbation types, indicating robust numerical decision boundaries rather than memorised edits. Robustness also transfers without target-domain data, significantly improving cross-lingual performance in Spanish. We further show that the same fine-tuning recipe confers robustness to evidence-side perturbations, using the VitaminC dataset.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 26 of its 32 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

26 of 32 references.

References · 26

  1. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  2. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,769
  3. 2023QLoRA: Efficient Finetuning of Quantized LLMscited by 3,355
  4. 2018FEVER: a large-scale dataset for Fact Extraction and VERificationcited by 1,362
  5. 2019Adversarial NLI: A New Benchmark for Natural Language Understandingcited by 798
  6. 2019Learning the Difference that Makes a Difference with Counterfactually-Augmented Datacited by 428
  7. 2021A Survey on Automated Fact-Checkingcited by 376
  8. 2024MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documentscited by 251
  9. 2021Gradient-based Adversarial Attacks against Text Transformerscited by 233
  10. 2021Get Your Vitamin C! Robust Fact Verification with Contrastive Evidencecited by 222
  11. 2021FEVEROUS: Fact Extraction and VERification Over Unstructured and Structured informationcited by 166
  12. 2019Do NLP Models Know Numbers? Probing Numeracy in Embeddingscited by 157
14 more
  1. 2023Fact-Checking Complex Claims with Program-Guided Reasoningcited by 146
  2. 2023AVeriTeC: A Dataset for Real-world Claim Verification with Evidence from the Webcited by 144
  3. 2020Injecting Numerical Reasoning Skills into Language Modelscited by 133
  4. 2023SCITAB: A Challenging Benchmark for Compositional Reasoning and Claim Verification on Scientific Tablescited by 51
  5. 2024MathOdyssey: Benchmarking Mathematical Problem-Solving Skills in Large Language Models Using Odyssey Math Datacited by 51
  6. 2023Exploring the Numerical Reasoning Capabilities of Language Models: A Comprehensive Analysis on Tabular Datacited by 33
  7. 2023MathAttack: Attacking Large Language Models Towards Math Solving Abilitycited by 31
  8. 2024NumeroLogic: Number Encoding for Enhanced LLMs' Numerical Reasoningcited by 31
  9. 2024QuanTemp: A real-world open-domain benchmark for fact-checking numerical claimscited by 22
  10. 2024Surprising Efficacy of Fine-Tuned Transformers for Fact-Checking over Larger Language Modelscited by 14
  11. 2024FREB-TQA: A Fine-Grained Robustness Evaluation Benchmark for Table Question Answeringcited by 14
  12. 2022Towards Robust Numerical Question Answering: Diagnosing Numerical Capabilities of NLP Systemscited by 6
  13. 2025NumPert: Numerical Perturbations to Probe Language Models for Veracity Predictioncited by 1
  14. 2026One LLM to Train Them All: Multi-Task Learning Framework for Fact-Checkingcited by 1

Built on by

Not cited yet.