mlsanityData to 7 Oct 2026

Helpful or Harmful Data? Fine-tuning-free Shapley Attribution for Explaining Language Model Predictions

arXiv 2406.04606 ↗First posted 7 Jun 2024cs.LG cs.AIcited by 13

Not on the map. Why?

Abstract (arXiv)

The increasing complexity of foundational models underscores the necessity for explainability, particularly for fine-tuning, the most widely used training method for adapting models to downstream tasks. Instance attribution, one type of explanation, attributes the model prediction to each training example by an instance score. However, the robustness of instance scores, specifically towards dataset resampling, has been overlooked. To bridge this gap, we propose a notion of robustness on the sign of the instance score. We theoretically and empirically demonstrate that the popular leave-one-out-based methods lack robustness, while the Shapley value behaves significantly better, but at a higher computational cost. Accordingly, we introduce an efficient fine-tuning-free approximation of the Shapley value (FreeShap) for instance attribution based on the neural tangent kernel. We empirically demonstrate that FreeShap outperforms other methods for instance attribution and other data-centric applications such as data removal, data selection, and wrong label detection, and further generalize our scale to large language models (LLMs). Our code is available at https://github.com/JTWang2000/FreeShap.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 55 of its 77 references that are in the map. First posted on arXiv 7 Jun 2024.

Builds on

55 of 77 references.

References · 55

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2023GPT-4 Technical Reportcited by 18,132
  4. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  5. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  6. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  7. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,040
  8. 2021On the Opportunities and Risks of Foundation Modelscited by 3,961
  9. 2017A Broad-Coverage Challenge Corpus for Sentence Understanding through Inferencecited by 2,665
  10. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
  11. 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
  12. 2020Making Pre-trained Language Models Better Few-shot Learnerscited by 1,255
38 more
  1. 2021GLM: General Language Model Pretraining with Autoregressive Blank Infillingcited by 1,026
  2. 2019Data Shapley: Equitable Valuation of Data for Machine Learningcited by 617
  3. 2019On Exact Computation with an Infinitely Wide Neural Netcited by 572
  4. 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 565
  5. 2024LESS: Selecting Influential Data for Targeted Instruction Tuningcited by 514
  6. 2018Towards Robust Interpretability with Self-Explaining Neural Networkscited by 505
  7. 2017Interpretation of Neural Networks is Fragilecited by 443
  8. 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 336
  9. 2023Studying Large Language Model Generalization with Influence Functionscited by 311
  10. 2019On the (In)fidelity and Sensitivity for Explanationscited by 309
  11. 2017Certified Defenses for Data Poisoning Attackscited by 288
  12. 2019Towards Efficient Data Valuation Based on the Shapley Valuecited by 279
  13. 2020GLISTER: Generalization based Data Subset Selection for Efficient and Robust Learningcited by 224
  14. 2015Visualizing and Understanding Neural Models in NLPcited by 199
  15. 2018Representer Point Selection for Explaining Deep Neural Networkscited by 187
  16. 2020Evaluating and Aggregating Feature-based Model Explanationscited by 185
  17. 2021Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine Learningcited by 162
  18. 2021Post-hoc Interpretability for Neural NLP: A Surveycited by 155
  19. 2020Explaining Black Box Predictions and Unveiling Data Artifacts through Influence Functionscited by 144
  20. 2022Data Banzhaf: A Robust Data Valuation Framework for Machine Learningcited by 126
  21. 2023DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Modelscited by 121
  22. 2020FastIF: Scalable Influence Functions for Efficient Model Interpretation and Debuggingcited by 120
  23. 2022A Kernel-Based View of Language Model Fine-Tuningcited by 118
  24. 2022More Than a Toy: Random Matrix Models Predict How Real-World Neural Representations Generalizecited by 87
  25. 2020Generating Hierarchical Explanations on Text Classification via Feature Interaction Detectioncited by 57
  26. 2013Bounding the Estimation Error of Sampling-based Shapley Value Approximationcited by 56
  27. 2021Explanation-Based Human Debugging of NLP Models: A Surveycited by 53
  28. 2021An Empirical Comparison of Instance Attribution Methods for NLPcited by 35
  29. 2022First is Better Than Last for Language Data Influencecited by 34
  30. 2023Approximating the Shapley Value without Marginal Contributionscited by 34
  31. 2024PINNACLE: PINN Adaptive ColLocation and Experimental points selectioncited by 31
  32. 2021Influence Tuning: Demoting Spurious Correlations via Instance Attribution and Instance-Driven Updatescited by 27
  33. 2022What Makes a Good Explanation?: A Harmonized View of Properties of Explanationscited by 21
  34. 2023Data Selection for Fine-tuning Large Language Models Using Transferred Shapley Valuescited by 21
  35. 2022Fooling Explanations in Text Classifierscited by 13
  36. 2023Source Attribution for Large Language Model-Generated Datacited by 11
  37. 2023Sample based Explanations via Generalized Representerscited by 10
  38. 2023Fair yet Asymptotically Equal Collaborative Learningcited by 8

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 9; 2026: 4.

13 papers.

Sort:

By area:

  1. 2025Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safetycited by 10
  2. 2025DUET: Optimizing Training Data Mixtures via Feedback from Unseen Evaluation Taskscited by 6
  3. 2025Group-robust Sample Reweighting for Subpopulation Shifts via Influence Functionscited by 6
  4. 2025Can Small Training Runs Reliably Guide Data Curation? Rethinking Proxy-Model Practicecited by 3
  5. 2025Unifying and Optimizing Data Values for Selection via Sequential Decision-Makingcited by 1
  6. 2025Interpret Policies in Deep Reinforcement Learning using SILVER with RL-Guided Labeling: A Model-level Approach to High-dimensional and Multi-action Environmentscited by 1
  7. 2025Shapley-based Data Valuation for LLM Alignment via Sequential Preference Optimizationcited by 1
  8. 2025Data Overvaluation Attack and Truthful Data Valuation in Federated Learningcited by 0
  9. 2025Uncovering Scaling Laws for Large Language Models via Inverse Problemscited by 0
  10. 2026The Chicken and Egg Dilemma: Co-optimizing Data and Model Configurations for LLMscited by 0
  11. 2026Is Data Shapley Not Better than Random in Data Selection? Ask NASHcited by 0
  12. 2026On the Fragility of Data Attribution When Learning Is Distributedcited by 0
1 more
  1. 2026De-attribute to Forget for LLM Unlearningcited by 0