pyDVL is a library of stable implementations of algorithms for data valuation and influence function computation
-
Updated
Sep 6, 2024 - Python
pyDVL is a library of stable implementations of algorithms for data valuation and influence function computation
Data-efficient Fine-tuning for LLM-based Recommendation (SIGIR'24)
Time series data contribution via influence functions
Influence Functions with (Eigenvalue-corrected) Kronecker-Factored Approximate Curvature
A simple PyTorch implementation of influence functions.
Supporting code for the paper "Finding Influential Training Samples for Gradient Boosted Decision Trees"
Influence Estimation for Gradient-Boosted Decision Trees
Official implementation of "Deeper Understanding of Black-box Predictions via Generalized Influence Functions".
👋 Influenciae is a Tensorflow Toolbox for Influence Functions
Intriguing Properties of Data Attribution on Diffusion Models (ICLR 2024)
This is a PyTorch reimplementation of Influence Functions from the ICML2017 best paper: Understanding Black-box Predictions via Influence Functions by Pang Wei Koh and Percy Liang.
This is an implementation of the paper ”Interpreting Twitter User Geolocation“.
This repo provides an implementation of the paper Interpreting Twitter User Geolocation.
[CVPR 2023] Regularizing Second-Order Influences for Continual Learning
A brief notebook on Influence Function (IF) for classical generative models (e.g., k-NN, KDE, GMM)
[EMNLP-2022 Findings] Code for paper “ProGen: Progressive Zero-shot Dataset Generation via In-context Feedback”.
An Empirical Study of Memorization in NLP (ACL 2022)
This is a [Stable] PyTorch reimplementation of Influence Functions from the ICML2017 best paper: Understanding Black-box Predictions via Influence Functions by Pang Wei Koh and Percy Liang.
Add a description, image, and links to the influence-functions topic page so that developers can more easily learn about it.
To associate your repository with the influence-functions topic, visit your repo's landing page and select "manage topics."