REVIEW 22 cited by
Datamodels: Predicting Predictions from Training Data
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
abstract
We present a conceptual framework, datamodeling, for analyzing the behavior of a model class in terms of the training data. For any fixed "target" example $x$, training set $S$, and learning algorithm, a datamodel is a parameterized function $2^S \to \mathbb{R}$ that for any subset of $S' \subset S$ -- using only information about which examples of $S$ are contained in $S'$ -- predicts the outcome of training a model on $S'$ and evaluating on $x$. Despite the potential complexity of the underlying process being approximated (e.g., end-to-end training and evaluation of deep neural networks), we show that even simple linear datamodels can successfully predict model outputs. We then demonstrate that datamodels give rise to a variety of applications, such as: accurately predicting the effect of dataset counterfactuals; identifying brittle predictions; finding semantically similar examples; quantifying train-test leakage; and embedding data into a well-behaved and feature-rich representation space. Data for this paper (including pre-computed datamodels as well as raw predictions from four million trained deep neural networks) is available at https://github.com/MadryLab/datamodels-data .
Forward citations
Cited by 22 Pith papers
-
MOON: Multi-Objective OrthoNormalized Updates for Multitask Learning
MOON applies spectral-nuclear-norm geometry to multi-objective gradient manipulation and uses polar-factor updates, with O(T^-1/2) deterministic and O(T^-1/4) stochastic convergence to Pareto stationarity.
-
Domain-Aware Scaling Laws Uncover Data Synergy
Domain-aware scaling laws with fitted γ and σ synergy terms recover stable code-math interactions from observational LLM mixtures and correctly predict mixture rankings in controlled small-scale trainings.
-
An Asymptotic Analysis of the Shapley Value for Dataset Valuation
Under smooth RKHS embedding utilities, a fixed owner's Shapley value is O(1/I)-close in L1 to an explicit leading term of scale (log I)/I driven by a first-order population signal.
-
Small edits, large models: How Wikipedia advocacy shapes LLM values
Wikipedia edits by animal welfare advocates measurably influence LLM outputs on animal welfare topics, shown via retrieval and gradient attribution plus fine-tuning experiments.
-
A Versatile Influence Function for Data Attribution with Non-Decomposable Loss
VIF approximates leave-one-out retraining influence for non-decomposable losses (contrastive, ranking, Cox) using a finite-difference of the loss evaluated via auto-differentiation.
-
A Cognac Shot To Forget Bad Memories: Corrective Unlearning for Graph Neural Networks
Corrective unlearning for graph neural networks is achieved by alternating contrastive separation of affected neighborhoods with asymmetric gradient ascent and descent, using as little as 5 percent of the manipulated set.
-
Understanding Data Influence with Differential Approximation
This paper introduces Diff-In, an influence estimator that accumulates second-order approximations of influence differences across training steps and shows strong accuracy in data cleaning, deletion, and coreset selec...
-
Better Training Data Attribution via Better Inverse Hessian-Vector Products
ASTRA, an EKFAC-preconditioned Neumann series iteration, computes more accurate inverse Hessian-vector products and improves training data attribution scores over EKFAC baselines.
-
AdaDeDup: Adaptive Hybrid Data Pruning for Efficient Large-Scale Object Detection Training
ADADEDUP adaptively prunes object detection datasets by combining semantic clustering with proxy-model loss feedback, matching full-data mAP at 20% pruning.
-
ClusterUCB: Efficient Gradient-Based Data Selection for Targeted Fine-Tuning of LLMs
ClusterUCB uses gradient clustering plus a modified UCB bandit to match full-budget gradient influence data selection at a 20% computing budget.
-
Daunce: Data Attribution through Uncertainty Estimation
DAUNCE computes training-data attribution as the covariance of per-example losses across an ensemble of perturbed fine-tuned models, reporting state-of-the-art LDS scores and the first attribution runs on proprietary LLMs.
-
Expert Survey: AI Reliability & Security Research Priorities
Expert ratings place capability forecasting and dangerous-capability evaluations at the top of a 105-area AI reliability and security research priority list.
-
Merge to Mix: Mixing Datasets via Model Merging
Merge to Mix shows that the performance of a parameter-averaged model predicts the performance of a model fine-tuned on any dataset mixture, enabling fast and accurate dataset mixture selection.
-
Towards Unified Attribution in Explainable AI, Data-Centric AI, and Mechanistic Interpretability
A position paper unifying feature, data, and component attribution under three shared techniques, perturbation, gradient, and linear approximation, and proposing cross-attribution research directions.
-
Improving Influence-based Instruction Tuning Data Selection for Balanced Learning of Diverse Capabilities
BIDS, a balanced influence-based data selection algorithm using per-task normalization and iterative greedy selection, improves balanced multi-capability instruction tuning and can outperform full-dataset training on ...
-
MoDoMoDo: Multi-Domain Data Mixtures for Multimodal LLM Reinforcement Learning
Multi-domain RLVR data mixing, guided by a quadratic surrogate fitted to 11 pilot runs, improves a Qwen2-VL-2B model's out-of-distribution accuracy by about 5 points over uniform mixing.
-
A Survey of LLM $\times$ DATA
A comprehensive survey of the bidirectional links between LLMs and data management, organized as DATA4LLM and LLM4DATA with a new 'IaaS' data-quality framework.
-
IDEAL: Data Equilibrium Adaptation for Multi-Capability Language Model Alignment
IDEAL tunes SFT data mixture proportions per domain with influence-function gradients, claiming about 7% average benchmark improvement over uniform mixing.
-
Capturing the Temporal Dependence of Training Data Influence
A new influence-estimation method approximates order-aware data influence with a dot product and reveals that early and late training data matter most.
-
Towards Data Governance of Frontier AI Models
Training data can serve as a governance lever for frontier AI through five proposed mechanisms: canary tokens, mandatory filtering, dataset reporting, data security, and know-your-customer rules.
-
What Is The Performance Ceiling of My Classifier? Utilizing Category-Wise Influence Functions for Pareto Frontier Analysis
Category-wise influence vectors plus linear programming and a genetic algorithm reweight training data to improve all classes at once, with an unproven criterion for when a classifier has reached its Pareto ceiling.
-
Scaling laws for activation steering with Llama 2 models and refusal mechanisms
Activation steering with contrastive vectors becomes less effective as Llama 2 models scale from 7B to 70B parameters, with peak effect at roughly 40% of the model's layers.
Discussion (0). Continue with ORCID to comment.