mlsanityData to 14 Oct 2026

HARDEN: Constrained Evolutionary Search for Harder, Answer-Preserving Evaluation Cases

arXiv 2609.30571 ↗First posted 24 Sep 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Language models are often evaluated on curated benchmarks that underrepresent the complexity of enterprise deployments. We introduce HARDEN, a constrained evolutionary search method to adapt the input of existing evaluation cases into more challenging variants while keeping their expected outputs fixed. HARDEN searches along generated domain-specific complexity axes while enforcing feasibility constraints such as preserving task semantics, realism, and execution validity. Across FinQA, PubMedQA, and ContractNLI and three Qwen3.5 model scales (35B-A3B, 122B-A10B, and 397B-A17B), HARDEN reduces task-model accuracy by 22.7% on average and by up to 49.9% relative to single-pass baselines using the same feasibility checks. These results show that evolutionary search can produce substantially harder valid evaluation cases.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 17 of its 25 references that are in the map. First posted on arXiv 24 Sep 2026.

Builds on

17 of 25 references.

References · 17

  1. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,148
  2. 2022Large Language Models Encode Clinical Knowledgecited by 1,830
  3. 2019PubMedQA: A Dataset for Biomedical Research Question Answeringcited by 1,288
  4. 2023Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generationcited by 819
  5. 2023Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLscited by 688
  6. 2017Adversarial Examples for Evaluating Reading Comprehension Systemscited by 571
  7. 2021FinQA: A Dataset of Numerical Reasoning over Financial Datacited by 562
  8. 2018Generating Natural Language Adversarial Examplescited by 424
  9. 2023LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Modelscited by 336
  10. 2023Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicinecited by 329
  11. 2024AdvPrompter: Fast Adaptive Adversarial Prompting for LLMscited by 156
  12. 2021ContractNLI: A Dataset for Document-level Natural Language Inference for Contractscited by 148
5 more
  1. 2022Stretching Sentence-pair NLI Models to Reason over Long Documents and Clusterscited by 44
  2. 2024DARG: Dynamic Evaluation of Large Language Models via Adaptive Reasoning Graphcited by 26
  3. 2025SECA: Semantically Equivalent and Coherent Attacks for Eliciting LLM Hallucinationscited by 9
  4. 2023Zero-Shot Question Answering over Financial Documents using Large Language Modelscited by 7
  5. 2026SQLMorph: Query Mutation and Fine-Grained Metrics for Text-to-SQL Evaluationcited by 1

Built on by

Not cited yet.