pytorch-normalizing-flows

Source

• Source kind: github-repository
• URL: https://github.com/karpathy/pytorch-normalizing-flows
• Discovery source: https://github.com/karpathy/pytorch-normalizing-flows
• License: NOASSERTION
• Distribution policy: public-summary-local-archive-only
• Public mirror status: summary-only
• Content hash: 74f42131399d62bc042d7e4ad638e60845a3cc0596a342f5d4faba1b73f0559d
• First seen: 2026-05-15
• Last changed: 2026-05-15

Classification

• Primary category: Minimal implementations
• Corpus source note: 2026-05-15-karpathy-public-corpus
• Project taxonomy: karpathy-project-taxonomy
• Idea map: karpathy-idea-map
• Topic hub: karpathy-public-work

Summary

pytorch-normalizing-flows Implementions of normalizing flows (NICE, RealNVP, MAF, IAF, Neural Splines Flows, etc) in PyTorch. !$1 todos - TODO: make work on GPU - TODO: 2D - ND: get (flat) using MNIST - TODO: ND - images (multi-scale architectures, Glow nets, etc) on MNIST/CIFAR/ImageNet - TODO: more stable residual-like IAF-style updates (tried but didn’t work too well) - TODO: parallel wavenet - TODO: radial/planar 2D flows from Rezende Mohamed 2015?

What This Teaches

• How a complex idea can be compressed into a minimal but working implementation.
• Useful as a reference style for serious small systems rather than decorative demos.

Why It Matters

This matters as part of Karpathy’s broader pattern: compress hard technical systems into readable, inspectable, working artifacts.

Repository Snapshot

• Full name: karpathy/pytorch-normalizing-flows
• Default branch: master
• HEAD: b60e119b37be10ce2930ef9fa17e58686aaf2b3d
• Stars at crawl: 915
• Forks at crawl: 102
• File count: 8
• README path: Readme.md
• License path: “
• Created: 2019-12-09T17:55:39Z
• Updated: 2026-04-25T13:29:22Z
• Pushed: 2020-01-27T18:37:10Z

Top-Level Structure

• nflib: 5
• [root]: 2
• assets: 1

File Extension Profile

• .py: 5
• .ipynb: 1
• .md: 1
• .png: 1

Tags / Release-Like Markers

• No git tags found in the shallow local clone.

Sample File Tree

• assets/moon_flow.png
• nflib/__init__.py
• nflib/flows.py
• nflib/made.py
• nflib/nets.py
• nflib/spline_flows.py
• nflib1.ipynb
• Readme.md

Public Handling Notes

• EXTRACTED: This page records public metadata and a source-grounded summary.
• INFERRED: Full local preservation, when available, is for private/local use unless a license or explicit source policy makes public redistribution safe.
• Do not treat this page as permission to republish unlicensed source text or code wholesale.