Brokttv/optimizers-from-scratch

training models with different optimizers using NumPy only. Featuring SGD, Adam, Adagrad, NAG, RMSProp, and Momentum. This repo also includes a benchmark against Pytorch developed optims.

28
/ 100
Experimental

This project offers efficient, fundamental implementations of deep learning optimizers like Adam, SGD, and RMSProp using NumPy. It allows machine learning practitioners to train models and understand how different optimization algorithms work under the hood. You input your model architecture and training data, and it outputs the optimized model parameters.

No commits in the last 6 months.

Use this if you are a machine learning researcher or student who needs to understand, customize, or benchmark core optimization algorithms for small-to-medium scale regression or classification tasks without relying on high-level frameworks.

Not ideal if you are working on large-scale deep learning projects that require GPU acceleration, advanced model architectures, or the extensive features of frameworks like PyTorch or TensorFlow.

machine-learning model-training algorithm-benchmarking numerical-optimization deep-learning-research
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 5 / 25
Maturity 15 / 25
Community 6 / 25

How are scores calculated?

Stars

13

Forks

1

Language

Jupyter Notebook

License

MIT

Last pushed

Sep 09, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Brokttv/optimizers-from-scratch"

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