mlsanityData to 7 Oct 2026

Atomic Consistency Preference Optimization for Long-Form Question Answering

arXiv 2505.09039 ↗First posted 14 May 2025cs.CLcited by 1

On the map: LLM alignment / RLHF / preference optimisation → LLM preference optimization methods

Abstract (arXiv)

Large Language Models (LLMs) often produce factoid hallucinations - plausible yet incorrect answers. A common mitigation strategy is model alignment, which improves factual accuracy by training on curated (factual, non-factual) pairs. However, this approach often relies on a stronger model (e.g., GPT-4) or an external knowledge base to assess factual correctness that may not always be accessible. Addressing this, we propose Atomic Consistency Preference Optimization (ACPO), a self-supervised preference-tuning method that enhances factual accuracy without external supervision. ACPO leverages atomic consistency signals (i.e., the agreement of individual facts across multiple stochastic responses) to identify high- and low-quality data pairs for model alignment. Despite being fully self-supervised, ACPO outperforms the strong supervised alignment baseline by 1.95 points averaged across Phi-3 and Llama3 on the LongFact and BioGen datasets, demonstrating its effectiveness in improving factual reliability without relying on external models or knowledge bases.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 27 of its 30 references that are in the map. First posted on arXiv 14 May 2025.

Builds on

27 of 30 references.

Closest ancestors · 4

Same direction, newest first

  1. 2024Self-Consistency Preference Optimization6 Nov
  2. 2024FactAlign: Long-form Factuality Alignment of Large Language Models2 Oct
  3. 2024Self-Alignment for Factuality: Mitigating Hallucinations in LLMs via Self-Evaluation14 Feb
  4. 2022Training language models to follow instructions with human feedback4 Mar

Other references · 23

  1. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  4. 2021SimCSE: Simple Contrastive Learning of Sentence Embeddingscited by 2,168
  5. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  6. 2023FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generationcited by 1,259
  7. 2023WizardLM: Empowering large pre-trained language models to follow complex instructionscited by 1,116
  8. 2023SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Modelscited by 787
  9. 2023RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedbackcited by 615
  10. 2023Chain-of-Verification Reduces Hallucination in Large Language Modelscited by 317
  11. 2023Evaluating Verifiability in Generative Search Enginescited by 297
  12. 2023Fine-tuning Language Models for Factualitycited by 219
11 more
  1. 2024The Effect of Sampling Temperature on Problem Solving in Large Language Modelscited by 168
  2. 2022A Review on Language Models as Knowledge Basescited by 156
  3. 2024Long-form factuality in large language modelscited by 146
  4. 2023The Troubling Emergence of Hallucination in Large Language Models -- An Extensive Definition, Quantification, and Prescriptive Remediationscited by 130
  5. 2023Do Language Models Know When They're Hallucinating References?cited by 115
  6. 2023Empower Large Language Model to Perform Better on Industrial Domain-Specific Question Answeringcited by 41
  7. 2023Halo: Estimation and Reduction of Hallucinations in Open-Source Weak Large Language Modelscited by 36
  8. 2024Calibrating Long-form Generations from Large Language Modelscited by 31
  9. 2020A Benchmark Dataset of Check-worthy Factual Claimscited by 27
  10. 2024Atomic Self-Consistency for Better Long Form Generationscited by 10
  11. 2024Exploring Language Model Generalization in Low-Resource Extractive QAcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Alignment Tuning for Large Language Models: A Data-Centric Lens on Alignment Data Pipelinescited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2024Beyond Under-Alignment: Atomic Preference Enhanced Factuality Tuning for Large Language Models
  2. 2024Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks
  3. 2025Latent Self-Consistency for Reliable Majority-Set Selection in Short- and Long-Answer Reasoning
  4. 2025InfoPO: On Mutual Information Maximization for Large Language Model Alignment
  5. 2024Anchored Preference Optimization and Contrastive Revisions: Addressing Underspecification in Alignment
  6. 2026Boosting Self-Consistency with Ranking
  7. 2025SSFO: Self-Supervised Faithfulness Optimization for Retrieval-Augmented Generation
  8. 2025Aligning Large Language Models for Faithful Integrity Against Opposing Argument