dfdazac/wassdistance
Approximating Wasserstein distances with PyTorch
This project helps data scientists and machine learning engineers understand the differences between complex data distributions, especially when dealing with high-dimensional datasets like images or text embeddings. It takes two sets of data points and calculates a 'distance' score, indicating how much they differ, which is useful for tasks like comparing generated data to real data. The primary users are researchers and practitioners working with advanced statistical modeling and generative models.
458 stars. No commits in the last 6 months.
Use this if you need to grasp the foundational concepts of Wasserstein distances and see practical examples of how they measure differences between probability distributions in a research or learning context.
Not ideal if you need a high-performance, GPU-optimized solution for calculating Wasserstein distances on large datasets, in which case a dedicated library like GeomLoss would be more suitable.
Stars
458
Forks
55
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Apr 15, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/dfdazac/wassdistance"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
PythonOT/POT
POT : Python Optimal Transport
recursionpharma/gflownet
GFlowNet library specialized for graph & molecular data
IShengFang/SpectralNormalizationKeras
Spectral Normalization for Keras Dense and Convolution Layers
iamalexkorotin/NeuralOptimalTransport
PyTorch implementation of "Neural Optimal Transport" (ICLR 2023 Spotlight)
PetrMokrov/Large-Scale-Wasserstein-Gradient-Flows
Source code for Large-Scale Wasserstein Gradient Flows (NeurIPS 2021)