mlsanityData to 7 Oct 2026

Factual Probing Is [MASK]: Learning vs. Learning to Recall

arXiv 2104.05240 ↗First posted 12 Apr 2021cs.CLcited by 291

Not on the map. Why?

Cited by 34 in Multimodal LLMs / vision-language models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Petroni et al. (2019) demonstrated that it is possible to retrieve world facts from a pre-trained language model by expressing them as cloze-style prompts and interpret the model's prediction accuracy as a lower bound on the amount of factual information it encodes. Subsequent work has attempted to tighten the estimate by searching for better prompts, using a disjoint set of facts as training data. In this work, we make two complementary contributions to better understand these factual probing techniques. First, we propose OptiPrompt, a novel and efficient method which directly optimizes in continuous embedding space. We find this simple method is able to predict an additional 6.4% of facts in the LAMA benchmark. Second, we raise a more important question: Can we really interpret these probing results as a lower bound? Is it possible that these prompt-search methods learn from the training data too? We find, somewhat surprisingly, that the training data used by these methods contains certain regularities of the underlying fact distribution, and all the existing prompt methods, including ours, are able to exploit them for better fact prediction. We conduct a set of control experiments to disentangle "learning" from "learning to recall", providing a more detailed picture of what different prompts can reveal about pre-trained language models.

Key facts. Cited by 291 papers in the map (data through 7 Oct 2026). Builds on 25 of its 33 references that are in the map. First posted on arXiv 12 Apr 2021.

Builds on

25 of 33 references.

References · 25

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  4. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
  5. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,878
  6. 2019ALBERT: A Lite BERT for Self-supervised Learning of Language Representationscited by 2,772
  7. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  8. 2019Language Models as Knowledge Bases?cited by 1,922
  9. 2020Making Pre-trained Language Models Better Few-shot Learnerscited by 1,255
  10. 2020RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Modelscited by 1,235
  11. 2019How Can We Know What Language Models Know?cited by 953
  12. 2020StereoSet: Measuring stereotypical bias in pretrained language modelscited by 911
13 more
  1. 2019Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inferencecited by 864
  2. 2020A Primer in BERTology: What we know about how BERT workscited by 807
  3. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 709
  4. 2020It's Not Just Size That Matters: Small Language Models Are Also Few-Shot Learnerscited by 668
  5. 2019Designing and Interpreting Probes with Control Taskscited by 608
  6. 2019What do you learn from context? Probing for sentence structure in contextualized word representationscited by 436
  7. 2020Information-Theoretic Probing with Minimum Description Lengthcited by 230
  8. 2020Information-Theoretic Probing for Linguistic Structurecited by 204
  9. 2017What do Neural Machine Translation Models Learn about Morphology?cited by 165
  10. 2019Open Sesame: Getting Inside BERT's Linguistic Knowledgecited by 119
  11. 2021BERTese: Learning to Speak to BERTcited by 82
  12. 2018What Makes Reading Comprehension Questions Easier?cited by 59
  13. 2020An information theoretic view on selecting linguistic probescited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 28; 2022: 76; 2023: 95; 2024: 74; 2025: 12; 2026: 6.

291 papers.

Sort:

By area:

  1. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  2. 2021Learning to Prompt for Vision-Language Modelscited by 2,269
  3. 2021Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processingcited by 2,222
  4. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  5. 2022Conditional Prompt Learning for Vision-Language Modelscited by 1,427
  6. 2021P-Tuning v2: Prompt Tuning Can Be Comparable to Fine-tuning Universally Across Scales and Taskscited by 686
  7. 2021Learning To Retrieve Prompts for In-Context Learningcited by 641
  8. 2021Recent Advances in Natural Language Processing via Large Pre-Trained Language Models: A Surveycited by 457
  9. 2023Adapting Language Models to Compress Contextscited by 350
  10. 2021True Few-Shot Learning with Language Modelscited by 343
  11. 2023Explainability for Large Language Models: A Surveycited by 331
  12. 2023MQuAKE: Assessing Knowledge Editing in Language Models via Multi-Hop Questionscited by 292
38 more
  1. 2022Exploring Visual Prompts for Adapting Large-Scale Modelscited by 271
  2. 2022Black-Box Tuning for Language-Model-as-a-Servicecited by 254
  3. 2023Think before you speak: Training Language Models With Pause Tokenscited by 254
  4. 2022Prompt Distribution Learningcited by 235
  5. 2022Data Contamination: From Memorization to Exploitationcited by 234
  6. 2021PPT: Pre-trained Prompt Tuning for Few-shot Learningcited by 232
  7. 2022Factuality Enhanced Language Models for Open-Ended Text Generationcited by 200
  8. 2021Noisy Channel Language Model Prompting for Few-Shot Text Classificationcited by 163
  9. 2022A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunitiescited by 162
  10. 2024A Comprehensive Study of Knowledge Editing for Large Language Modelscited by 157
  11. 2022A Review on Language Models as Knowledge Basescited by 156
  12. 2023LM vs LM: Detecting Factual Errors via Cross Examinationcited by 155
  13. 2023Active Prompting with Chain-of-Thought for Large Language Modelscited by 154
  14. 2023All in One: Multi-task Prompting for Graph Neural Networkscited by 147
  15. 2021Cutting Down on Prompts and Parameters: Simple Few-Shot Learning with Language Modelscited by 139
  16. 2022Neural Prompt Searchcited by 133
  17. 2022Domain Adaptation via Prompt Learningcited by 128
  18. 2022Unsupervised Prompt Learning for Vision-Language Modelscited by 125
  19. 2022Natural Language to Code Translation with Executioncited by 122
  20. 2023Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Datacited by 119
  21. 2021Differentiable Prompt Makes Pre-trained Language Models Better Few-shot Learnerscited by 115
  22. 2023The Earth is Flat because...: Investigating LLMs' Belief towards Misinformation via Persuasive Conversationcited by 107
  23. 2023Text Classification via Large Language Modelscited by 104
  24. 2023Generating Benchmarks for Factuality Evaluation of Language Modelscited by 103
  25. 2022SimKGC: Simple Contrastive Knowledge Graph Completion with Pre-trained Language Modelscited by 90
  26. 2024LISA: Layerwise Importance Sampling for Memory-Efficient Large Language Model Fine-Tuningcited by 88
  27. 2021Can Language Models be Biomedical Knowledge Bases?cited by 83
  28. 2021Why Do Pretrained Language Models Help in Downstream Tasks? An Analysis of Head and Prompt Tuningcited by 82
  29. 2023Modular Deep Learningcited by 82
  30. 2022Revisiting Parameter-Efficient Tuning: Are We Really There Yet?cited by 76
  31. 2022PromptBERT: Improving BERT Sentence Embeddings with Promptscited by 75
  32. 2022Iteratively Prompt Pre-trained Language Models for Chain of Thoughtcited by 75
  33. 2021On Transferability of Prompt Tuning for Natural Language Processingcited by 74
  34. 2023Towards Efficient Visual Adaption via Structural Re-parameterizationcited by 73
  35. 2023Large Language Models and Knowledge Graphs: Opportunities and Challengescited by 71
  36. 2022Black-box Prompt Learning for Pre-trained Language Modelscited by 70
  37. 2023A Closer Look at the Few-Shot Adaptation of Large Vision-Language Modelscited by 67
  38. 2022Finding Skill Neurons in Pre-trained Transformer-based Language Modelscited by 65

50 most cited of 291 All 291 (JSON)