My optimization(With and Without gradient) library

Deep Neural Network Optimization Platform with Gradient-based, Gradient-Free Algorithms

Performing Dense correspondence matching and Depth Image Generation using Patch Matching and Gradients-Based Features algorithms.

PyTorch implementation of influence functions: ICML 2017 method, TracIn (NeurIPS 2020) and EmpiricalIF (NeurIPS 2022). Estimate how each training sample affects model predictions without retraining.

Python code for the paper “Enhanced gradient-based MCMC in discrete spaces”, TMLR 2022, https://openreview.net/forum?id=j2Mid5hFUJ