mlsanityData to 14 Oct 2026

Rich Knowledge Sources Bring Complex Knowledge Conflicts: Recalibrating Models to Reflect Conflicting Evidence

arXiv 2210.13701 ↗First posted 25 Oct 2022cs.CLcited by 144

Not on the map. Why?

Cited by 41 in Retrieval-augmented generation, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Question answering models can use rich knowledge sources -- up to one hundred retrieved passages and parametric knowledge in the large-scale language model (LM). Prior work assumes information in such knowledge sources is consistent with each other, paying little attention to how models blend information stored in their LM parameters with that from retrieved evidence documents. In this paper, we simulate knowledge conflicts (i.e., where parametric knowledge suggests one answer and different passages suggest different answers) and examine model behaviors. We find retrieval performance heavily impacts which sources models rely on, and current models mostly rely on non-parametric knowledge in their best-performing settings. We discover a troubling trend that contradictions among knowledge sources affect model confidence only marginally. To address this issue, we present a new calibration study, where models are discouraged from presenting any single answer when presented with multiple conflicting answer candidates in retrieved evidences.

Key facts. Cited by 144 papers in the map (data through 14 Oct 2026). Builds on 41 of its 51 references that are in the map. First posted on arXiv 25 Oct 2022.

Builds on

41 of 51 references.

References · 41

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  3. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  4. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  5. 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,166
  6. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  7. 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,824
  8. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  9. 2020REALM: Retrieval-Augmented Language Model Pre-Trainingcited by 2,141
  10. 2019Language Models as Knowledge Bases?cited by 1,936
  11. 2022LaMDA: Language Models for Dialog Applicationscited by 1,220
  12. 2020Leveraging Passage Retrieval with Generative Models for Open Domain Question Answeringcited by 1,210
29 more
  1. 2021Scaling Language Models: Methods, Analysis & Insights from Training Gophercited by 1,202
  2. 2020Beyond Accuracy: Behavioral Testing of NLP models with CheckListcited by 1,006
  3. 2017Reading Wikipedia to Answer Open-Domain Questionscited by 777
  4. 2020How Much Knowledge Can You Pack Into the Parameters of a Language Model?cited by 719
  5. 2019Latent Retrieval for Weakly Supervised Open Domain Question Answeringcited by 695
  6. 2019ELI5: Long Form Question Answeringcited by 575
  7. 2017Adversarial Examples for Evaluating Reading Comprehension Systemscited by 571
  8. 2019Learning the Difference that Makes a Difference with Counterfactually-Augmented Datacited by 439
  9. 2020AmbigQA: Answering Ambiguous Open-domain Questionscited by 390
  10. 2021Entity-Based Knowledge Conflicts in Question Answeringcited by 356
  11. 2016Bidirectional Attention Flow for Machine Comprehensioncited by 278
  12. 2020Selective Question Answering under Domain Shiftcited by 238
  13. 2020Distilling Knowledge from Reader to Retriever for Question Answeringcited by 226
  14. 2021Polyjuice: Generating Counterfactuals for Explaining, Evaluating, and Improving Modelscited by 222
  15. 2019MRQA 2019 Shared Task: Evaluating Generalization in Reading Comprehensioncited by 184
  16. 2021Hurdles to Progress in Long-form Question Answeringcited by 167
  17. 2021SituatedQA: Incorporating Extra-Linguistic Contexts into QAcited by 145
  18. 2020Question and Answer Test-Train Overlap in Open-Domain Question Answering Datasetscited by 138
  19. 2021End-to-End Training of Multi-Document Reader and Retriever for Open-Domain Question Answeringcited by 133
  20. 2020Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehensioncited by 131
  21. 2020Enabling Language Models to Fill in the Blankscited by 122
  22. 2019Seeing Things from a Different Angle: Discovering Diverse Perspectives about Claimscited by 74
  23. 2021Knowing More About Questions Can Help: Improving Calibration in Question Answeringcited by 47
  24. 2019Universal Dependency Parsing from Scratchcited by 46
  25. 2022Synthetic Disinformation Attacks on Automated Fact Verification Systemscited by 44
  26. 2020Is Retriever Merely an Approximator of Reader?cited by 30
  27. 2021Retrieval-guided Counterfactual Generation for QAcited by 29
  28. 2019Automatic Fact-guided Sentence Modificationcited by 28
  29. 2022Reasoning over Logically Interacted Conditions for Question Answeringcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 3; 2023: 19; 2024: 46; 2025: 35; 2026: 41.

144 papers.

Sort:

By area:

  1. 2022When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memoriescited by 1,128
  2. 2023Making Retrieval-Augmented Language Models Robust to Irrelevant Contextcited by 369
  3. 2023Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflictscited by 341
  4. 2023Trusting Your Evidence: Hallucinate Less with Context-aware Decodingcited by 327
  5. 2024Knowledge Conflicts for LLMs: A Surveycited by 283
  6. 2023Fine-tuning Language Models for Factualitycited by 221
  7. 2023Survey on Factuality in Large Language Models: Knowledge, Retrieval and Domain-Specificitycited by 206
  8. 2024Retrieval-Augmented Generation with Graphs (GraphRAG)cited by 172
  9. 2023On the Risk of Misinformation Pollution with Large Language Modelscited by 128
  10. 2022DisentQA: Disentangling Parametric and Contextual Knowledge with Counterfactual Question Answeringcited by 96
  11. 2024Tug-of-War Between Knowledge: Exploring and Resolving Knowledge Conflicts in Retrieval-Augmented Language Modelscited by 93
  12. 2024Cutting Off the Head Ends the Conflict: A Mechanism for Interpreting and Mitigating Knowledge Conflicts in Language Modelscited by 91
38 more
  1. 2024FaithEval: Can Your Language Model Stay Faithful to Context, Even If "The Moon is Made of Marshmallows"cited by 86
  2. 2024What Evidence Do Language Models Find Convincing?cited by 75
  3. 2025SealQA: Raising the Bar for Reasoning in Search-Augmented Language Modelscited by 75
  4. 2024Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineeringcited by 63
  5. 2025Retrieval-Augmented Generation with Conflicting Evidencecited by 63
  6. 2023Why So Gullible? Enhancing the Robustness of Retrieval-Augmented Models against Counterfactual Noisecited by 54
  7. 2023Resolving Knowledge Conflicts in Large Language Modelscited by 50
  8. 2024Context-DPO: Aligning Language Models for Context-Faithfulnesscited by 50
  9. 2024Enhancing Contextual Understanding in Large Language Models through Contrastive Decodingcited by 49
  10. 2024Blinded by Generated Contexts: How Language Models Merge Generated and Retrieved Contexts When Knowledge Conflicts?cited by 48
  11. 2023Personas as a Way to Model Truthfulness in Language Modelscited by 47
  12. 2023Understanding Retrieval Augmentation for Long-Form Question Answeringcited by 45
  13. 2024WikiContradict: A Benchmark for Evaluating LLMs on Real-World Knowledge Conflicts from Wikipediacited by 44
  14. 2024IRCAN: Mitigating Knowledge Conflicts in LLM Generation via Identifying and Reweighting Context-Aware Neuronscited by 38
  15. 2024Insight Over Sight: Exploring the Vision-Knowledge Conflicts in Multimodal LLMscited by 38
  16. 2024AdaCAD: Adaptively Decoding to Balance Conflicts between Contextual and Parametric Knowledgecited by 37
  17. 2023Merging Generated and Retrieved Knowledge for Open-Domain QAcited by 36
  18. 2023Towards Optimizing with Large Language Modelscited by 35
  19. 2023How Do Large Language Models Capture the Ever-changing World Knowledge? A Review of Recent Advancescited by 33
  20. 2024RAG-DDR: Optimizing Retrieval-Augmented Generation Using Differentiable Data Rewardscited by 33
  21. 2025Taming Knowledge Conflicts in Language Modelscited by 30
  22. 2023Trends in Integration of Knowledge and Large Language Models: A Survey and Taxonomy of Methods, Benchmarks, and Applicationscited by 28
  23. 2024Discerning and Resolving Knowledge Conflicts through Adaptive Decoding with Contextual Information-Entropy Constraintcited by 27
  24. 2024Studying Large Language Model Behaviors Under Context-Memory Conflicts With Real Documentscited by 27
  25. 2024Context versus Prior Knowledge in Language Modelscited by 26
  26. 2024Claim Verification in the Age of Large Language Models: A Surveycited by 24
  27. 2023Intuitive or Dependent? Investigating LLMs' Behavior Style to Conflicting Promptscited by 23
  28. 2023A Causal View of Entity Bias in (Large) Language Modelscited by 21
  29. 2023Mitigating Temporal Misalignment by Discarding Outdated Factscited by 21
  30. 2024AmbigDocs: Reasoning across Documents on Different Entities under the Same Namecited by 21
  31. 2024KnowPO: Knowledge-aware Preference Optimization for Controllable Knowledge Selection in Retrieval-Augmented Language Modelscited by 20
  32. 2025Benchmarking Multimodal Knowledge Conflict for Large Multimodal Modelscited by 19
  33. 2024ECon: On the Detection and Resolution of Evidence Conflictscited by 18
  34. 2024Open Domain Question Answering with Conflicting Contextscited by 18
  35. 2025Robust Multimodal Large Language Models Against Modality Conflictcited by 18
  36. 2025ReliabilityRAG: Effective and Provably Robust Defense for RAG-based Web-Searchcited by 18
  37. 2024Is Factuality Enhancement a Free Lunch For LLMs? Better Factuality Can Lead to Worse Context-Faithfulnesscited by 17
  38. 2025Task Matters: Knowledge Requirements Shape LLM Responses to Context-Memory Conflictcited by 16

50 most cited of 144 All 144 (JSON)