Related papers: In-Context Probing Approximates Influence Function for Data Valuation

In-Context Probing Approximates Influence Function for Data Valuation

URL: http://arxiv.org/abs/2407.12259v1
Date: Wed, 17 Jul 2024 02:06:56 GMT
Title: In-Context Probing Approximates Influence Function for Data Valuation
Authors: Cathy Jiao, Gary Gao, Chenyan Xiong,
Abstract summary: We show that data valuation through in-context probing approximates influence functions for selecting training data. Our empirical findings show that in-context probing and gradient-based influence frameworks are similar in how they rank training data.
Score: 16.404477234171733
License: http://creativecommons.org/licenses/by/4.0/
Abstract: Data valuation quantifies the value of training data, and is used for data attribution (i.e., determining the contribution of training data towards model predictions), and data selection; both of which are important for curating high-quality datasets to train large language models. In our paper, we show that data valuation through in-context probing (i.e., prompting a LLM) approximates influence functions for selecting training data. We provide a theoretical sketch on this connection based on transformer models performing "implicit" gradient descent on its in-context inputs. Our empirical findings show that in-context probing and gradient-based influence frameworks are similar in how they rank training data. Furthermore, fine-tuning experiments on data selected by either method reveal similar model performance.

Related papers

Capturing the Temporal Dependence of Training Data Influence [100.91355498124527]
We formalize the concept of trajectory-specific leave-one-out influence, which quantifies the impact of removing a data point during training. We propose data value embedding, a novel technique enabling efficient approximation of trajectory-specific LOO. As data value embedding captures training data ordering, it offers valuable insights into model training dynamics.
arXiv Detail & Related papers (2024-12-12T18:28:55Z)
Dataset Distillation-based Hybrid Federated Learning on Non-IID Data [19.01147151081893]
We propose a hybrid federated learning framework called HFLDD, which integrates dataset distillation to generate independent and equally distributed (IID) data. We partition the clients into heterogeneous clusters, where the data labels among different clients within a cluster are unbalanced. This training process is like traditional federated learning on IID data, and hence effectively alleviates the impact of Non-IID data on model training.
arXiv Detail & Related papers (2024-09-26T03:52:41Z)
Entropy Law: The Story Behind Data Compression and LLM Performance [115.70395740286422]
We find that model performance is negatively correlated to the compression ratio of training data, which usually yields a lower training loss. Based on the findings of the entropy law, we propose a quite efficient and universal data selection method. We also present an interesting application of entropy law that can detect potential performance risks at the beginning of model training.
arXiv Detail & Related papers (2024-07-09T08:14:29Z)
A Universal Metric of Dataset Similarity for Cross-silo Federated Learning [0.0]
Federated learning is increasingly used in domains such as healthcare to facilitate model training without data-sharing. In this paper, we propose a novel metric for assessing dataset similarity. We show that our metric shows a robust and interpretable relationship with model performance and can be calculated in privacy-preserving manner.
arXiv Detail & Related papers (2024-04-29T15:08:24Z)
LESS: Selecting Influential Data for Targeted Instruction Tuning [64.78894228923619]
We propose LESS, an efficient algorithm to estimate data influences and perform Low-rank gradiEnt Similarity Search for instruction data selection. We show that training on a LESS-selected 5% of the data can often outperform training on the full dataset across diverse downstream tasks. Our method goes beyond surface form cues to identify data that the necessary reasoning skills for the intended downstream application.
arXiv Detail & Related papers (2024-02-06T19:18:04Z)
Stubborn Lexical Bias in Data and Models [50.79738900885665]
We use a new statistical method to examine whether spurious patterns in data appear in models trained on the data. We apply an optimization approach to *reweight* the training data, reducing thousands of spurious correlations. Surprisingly, though this method can successfully reduce lexical biases in the training data, we still find strong evidence of corresponding bias in the trained models.
arXiv Detail & Related papers (2023-06-03T20:12:27Z)
Data-SUITE: Data-centric identification of in-distribution incongruous examples [81.21462458089142]
Data-SUITE is a data-centric framework to identify incongruous regions of in-distribution (ID) data. We empirically validate Data-SUITE's performance and coverage guarantees.
arXiv Detail & Related papers (2022-02-17T18:58:31Z)
Combining Feature and Instance Attribution to Detect Artifacts [62.63504976810927]
We propose methods to facilitate identification of training data artifacts. We show that this proposed training-feature attribution approach can be used to uncover artifacts in training data. We execute a small user study to evaluate whether these methods are useful to NLP researchers in practice.
arXiv Detail & Related papers (2021-07-01T09:26:13Z)
Deep Stable Learning for Out-Of-Distribution Generalization [27.437046504902938]
Approaches based on deep neural networks have achieved striking performance when testing data and training data share similar distribution. Eliminating the impact of distribution shifts between training and testing data is crucial for building performance-promising deep models. We propose to address this problem by removing the dependencies between features via learning weights for training samples.
arXiv Detail & Related papers (2021-04-16T03:54:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.