MarioSieg/magnetron

(WIP) A small but powerful, homemade PyTorch from scratch.

56
/ 100
Established

This is a machine learning runtime designed for developers who build and optimize custom ML solutions. It allows you to feed in your model architecture and data to train or run inference, producing trained models or predictions. It's for engineers and researchers who need granular control over the full ML stack, from execution to memory.

676 stars. Actively maintained with 10 commits in the last 30 days.

Use this if you are a machine learning systems engineer or researcher who needs deep control over how your models run, experiments with new kernels or execution strategies, or ports ML workloads to specialized hardware.

Not ideal if you are looking for a high-level framework with a vast ecosystem, extensive pre-built features, and abstraction layers for rapid application development without needing to dive into implementation details.

machine-learning-engineering ML-systems-design custom-hardware-ML performance-optimization deep-learning-research
No Package No Dependents
Maintenance 17 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 13 / 25

How are scores calculated?

Stars

676

Forks

33

Language

C++

License

Last pushed

Mar 11, 2026

Commits (30d)

10

Get this data via API

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

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