mlsanityData to 7 Oct 2026

Studying Large Language Model Generalization with Influence Functions

arXiv 2308.03296 ↗First posted 7 Aug 2023cs.LG cs.CL stat.MLcited by 311

Not on the map. Why?

Cited by 27 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of evidence is: which training examples most contribute to a given behavior? Influence functions aim to answer a counterfactual: how would the model's parameters (and hence its outputs) change if a given sequence were added to the training set? While influence functions have produced insights for small models, they are difficult to scale to large language models (LLMs) due to the difficulty of computing an inverse-Hessian-vector product (IHVP). We use the Eigenvalue-corrected Kronecker-Factored Approximate Curvature (EK-FAC) approximation to scale influence functions up to LLMs with up to 52 billion parameters. In our experiments, EK-FAC achieves similar accuracy to traditional influence function estimators despite the IHVP computation being orders of magnitude faster. We investigate two algorithmic techniques to reduce the cost of computing gradients of candidate training sequences: TF-IDF filtering and query batching. We use influence functions to investigate the generalization patterns of LLMs, including the sparsity of the influence patterns, increasing abstraction with scale, math and programming abilities, cross-lingual generalization, and role-playing behavior. Despite many apparently sophisticated forms of generalization, we identify a surprising limitation: influences decay to near-zero when the order of key phrases is flipped. Overall, influence functions give us a powerful new tool for studying the generalization properties of LLMs.

Key facts. Cited by 311 papers in the map (data through 7 Oct 2026). First posted on arXiv 7 Aug 2023.

Builds on

0 of 0 references.

No references here yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 25; 2024: 92; 2025: 111; 2026: 83.

311 papers.

Sort:

By area:

  1. 2023The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"cited by 399
  2. 2023SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generationcited by 376
  3. 2023Explainability for Large Language Models: A Surveycited by 331
  4. 2024Knowledge Conflicts for LLMs: A Surveycited by 274
  5. 2024Rethinking Machine Unlearning for Large Language Modelscited by 247
  6. 2024Large Language Model-Brained GUI Agents: A Surveycited by 160
  7. 2023Taken out of context: On measuring situational awareness in LLMscited by 131
  8. 2023DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Modelscited by 121
  9. 2024Rethinking Interpretability in the Era of Large Language Modelscited by 110
  10. 2024SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearningcited by 104
  11. 2024Bias and Unfairness in Information Retrieval Systems: New Challenges in the LLM Eracited by 97
  12. 2024MATES: Model-Aware Data Selection for Efficient Pretraining with Data Influence Modelscited by 96
38 more
  1. 2024ContextCite: Attributing Model Generation to Contextcited by 93
  2. 2024What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functionscited by 88
  3. 2024A Primer on the Inner Workings of Transformer-based Language Modelscited by 83
  4. 2024Data Shapley in One Training Runcited by 76
  5. 2024Reliable, Adaptable, and Attributable Language Models with Retrievalcited by 73
  6. 2024MoRAL: MoE Augmented LoRA for LLMs' Lifelong Learningcited by 63
  7. 2024SelfIE: Self-Interpretation of Large Language Model Embeddingscited by 63
  8. 2023Kronecker-Factored Approximate Curvature for Modern Neural Network Architecturescited by 49
  9. 2023Intriguing Properties of Data Attribution on Diffusion Modelscited by 47
  10. 2024Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Eracited by 45
  11. 2024A Language Model's Guide Through Latent Spacecited by 43
  12. 2025The State of Multilingual LLM Safety Research: From Measuring the Language Gap to Mitigating Itcited by 42
  13. 2024Procedural Knowledge in Pretraining Drives Reasoning in Large Language Modelscited by 41
  14. 2024Scalable Influence and Fact Tracing for Large Language Model Pretrainingcited by 40
  15. 2024Summing Up the Facts: Additive Mechanisms Behind Factual Recall in LLMscited by 39
  16. 2024Data Selection via Optimal Control for Language Modelscited by 38
  17. 2025LLM Unlearning via Neural Activation Redirectioncited by 38
  18. 2023Merging by Matching Models in Task Parameter Subspacescited by 36
  19. 2023Evaluating the Reversal Curse in Model Editingcited by 34
  20. 2024Harnessing Diversity for Important Data Selection in Pretraining Large Language Modelscited by 34
  21. 2024Capturing the Temporal Dependence of Training Data Influencecited by 34
  22. 2025Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuningcited by 34
  23. 2024Most Influential Subset Selection: Challenges, Promises, and Beyondcited by 31
  24. 2024Influence Functions for Scalable Data Attribution in Diffusion Modelscited by 31
  25. 2024Source-Aware Training Enables Knowledge Attribution in Language Modelscited by 29
  26. 2024Unleashing the Power of Data Tsunami: A Comprehensive Survey on Data Assessment and Selection for Instruction Tuning of Language Modelscited by 29
  27. 2024Not All LLM Reasoners Are Created Equalcited by 29
  28. 2024Unraveling Cross-Modality Knowledge Conflicts in Large Vision-Language Modelscited by 28
  29. 2024On Effects of Steering Latent Representation for Large Language Model Unlearningcited by 27
  30. 2024Corrective Machine Unlearningcited by 26
  31. 2024WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Modelscited by 25
  32. 2024Large Language Models in Drug Discovery and Development: From Disease Mechanisms to Clinical Trialscited by 24
  33. 2024Towards Understanding Multi-Task Learning (Generalization) of LLMs via Detecting and Exploring Task-Specific Neuronscited by 23
  34. 2024AttriBoT: A Bag of Tricks for Efficiently Approximating Leave-One-Out Context Attributioncited by 23
  35. 2025Learning to Optimize Multi-Objective Alignment Through Dynamic Reward Weightingcited by 23
  36. 2024Explainable Generative AI (GenXAI): A Survey, Conceptualization, and Research Agendacited by 22
  37. 2024Data Attribution for Text-to-Image Models by Unlearning Synthesized Imagescited by 22
  38. 2024What Do Learning Dynamics Reveal About Generalization in LLM Reasoning?cited by 22

50 most cited of 311 All 311 (JSON)