aliutkus/groupmap

GroupMap: beyond mean and variance matching for deep learning

21
/ 100
Experimental

This project offers an alternative to traditional normalization methods in deep learning, transforming input data so it precisely matches a user-defined distribution like uniform, Gaussian, or Cauchy. Instead of just adjusting the mean and variance, it remaps the entire data distribution. This is intended for deep learning researchers and practitioners who need fine-grained control over the statistical properties of their model's internal representations.

No commits in the last 6 months.

Use this if you need to enforce a specific arbitrary statistical distribution on your deep learning model's activations or input features, rather than just normalizing their mean and variance.

Not ideal if you require your model to track and apply statistics learned over an entire training dataset during inference, as this implementation always computes statistics from the current batch.

deep-learning-research neural-network-training data-distribution-control machine-learning-engineering
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

11

Forks

Language

Python

License

MIT

Last pushed

Sep 27, 2022

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/aliutkus/groupmap"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.