mlsanityData to 7 Oct 2026

Do Influence Functions Work on Large Language Models?

arXiv 2409.19998 ↗First posted 30 Sep 2024cs.CL cs.AIcited by 20

Not on the map. Why?

Abstract (arXiv)

Influence functions are important for quantifying the impact of individual training data points on a model's predictions. Although extensive research has been conducted on influence functions in traditional machine learning models, their application to large language models (LLMs) has been limited. In this work, we conduct a systematic study to address a key question: do influence functions work on LLMs? Specifically, we evaluate influence functions across multiple tasks and find that they consistently perform poorly in most settings. Our further investigation reveals that their poor performance can be attributed to: (1) inevitable approximation errors when estimating the iHVP component due to the scale of LLMs, (2) uncertain convergence during fine-tuning, and, more fundamentally, (3) the definition itself, as changes in model parameters do not necessarily correlate with changes in LLM behavior. Thus, our study suggests the need for alternative approaches for identifying influential samples.

Key facts. Cited by 20 papers in the map (data through 7 Oct 2026). Builds on 39 of its 59 references that are in the map. First posted on arXiv 30 Sep 2024.

Builds on

39 of 59 references.

References · 39

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2023GPT-4 Technical Reportcited by 18,132
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2024The Llama 3 Herd of Modelscited by 14,192
  5. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  6. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  7. 2020Scaling Laws for Neural Language Modelscited by 6,411
  8. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  9. 2023QLoRA: Efficient Finetuning of Quantized LLMscited by 3,355
  10. 2023Mistral 7Bcited by 3,065
  11. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 2,914
  12. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,250
27 more
  1. 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
  2. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  3. 2017Visualizing the Loss Landscape of Neural Netscited by 1,126
  4. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,106
  5. 2023LIMA: Less Is More for Alignmentcited by 1,080
  6. 2023A Simple and Effective Pruning Approach for Large Language Modelscited by 776
  7. 2019Data Shapley: Equitable Valuation of Data for Machine Learningcited by 617
  8. 2024TinyLlama: An Open-Source Small Language Modelcited by 604
  9. 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 565
  10. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingcited by 478
  11. 2023Scaling Data-Constrained Language Modelscited by 391
  12. 2023Studying Large Language Model Generalization with Influence Functionscited by 311
  13. 2023TRAK: Attributing Model Behavior at Scalecited by 273
  14. 2021Scaling Laws for Transfercited by 265
  15. 2023LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70Bcited by 183
  16. 2021Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learningcited by 162
  17. 2022If Influence Functions are the Answer, Then What is the Question?cited by 142
  18. 2021Scaling Up Influence Functionscited by 130
  19. 2023DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Modelscited by 121
  20. 2020FastIF: Scalable Influence Functions for Efficient Model Interpretation and Debuggingcited by 120
  21. 2023Universal Jailbreak Backdoors from Poisoned Human Feedbackcited by 120
  22. 2024What is in Your Safe Data? Identifying Benign Data that Breaks Safetycited by 114
  23. 2024What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functionscited by 88
  24. 2021Input Similarity from the Neural Network Perspectivecited by 78
  25. 2024BEEAR: Embedding-based Adversarial Removal of Safety Backdoors in Instruction-tuned Language Modelscited by 50
  26. 2023Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injectionscited by 37
  27. 2024Token-wise Influential Training Data Retrieval for Large Language Modelscited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 9; 2026: 11.

20 papers.

Sort:

By area:

  1. 2026Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Modelscited by 15
  2. 2025Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Schedulercited by 12
  3. 2025Better Training Data Attribution via Better Inverse Hessian-Vector Productscited by 10
  4. 2025IF-GUIDE: Influence Function-Guided Detoxification of LLMscited by 7
  5. 2025Fast Data Attribution for Text-to-Image Modelscited by 7
  6. 2026Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attributioncited by 6
  7. 2025First is Not Really Better Than Last: Evaluating Layer Choice and Aggregation Strategies in Language Model Data Influence Estimationcited by 5
  8. 2025Scalable Valuation of Human Feedback through Provably Robust Model Alignmentcited by 4
  9. 2025Where Did It Go Wrong? Attributing Undesirable LLM Behaviors via Representation Gradient Tracingcited by 4
  10. 2025Causally-Enhanced Reinforcement Policy Optimizationcited by 2
  11. 2025Z0-Inf: Zeroth Order Approximation for Data Influencecited by 2
  12. 2026An Isotropic Approach to Efficient Uncertainty Quantification with Gradient Normscited by 1
8 more
  1. 2026DRIFT: Data Selection for LLM Instruction Tuning via On-Policy Attributioncited by 1
  2. 2026Divide and Predict: An Architecture for Input Space Partitioning and Enhanced Accuracycited by 0
  3. 2026GradShield: Alignment Preserving Finetuningcited by 0
  4. 2026Validity Threats for Foundation Model Researchcited by 0
  5. 2026Influcoder: Distilling Decoders' Gradient Influence Rankings into an Encoder for Data Attributioncited by 0
  6. 2026From Reweighting to Rewriting: Unlocking the Intervention Effects of Influential Samples in Training Data Attributioncited by 0
  7. 2026Can Data Attribution Filter Out Subliminal Learning? Not Reliablycited by 0
  8. 2026Which Influence Are We Estimating? The Role of Counterfactual Specifications in Data Attributioncited by 0