BobMcDear/attorch

A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.

41
/ 100
Emerging

This project offers a collection of neural network building blocks, like convolution layers and activation functions, that are designed to be easily customizable and highly efficient. It takes in raw data or intermediate feature maps and processes them through optimized network layers, outputting improved feature representations or predictions. This is for machine learning engineers and researchers who build and train deep learning models and need fine-grained control over performance.

597 stars. No commits in the last 6 months.

Use this if you are developing custom deep learning operations and want to create fast, specialized neural network components without writing complex CUDA code.

Not ideal if you primarily need off-the-shelf convolutional or pooling layers, as PyTorch's native implementations are generally faster for these specific operations.

deep-learning-engineering neural-network-design model-optimization custom-operations machine-learning-research
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 13 / 25

How are scores calculated?

Stars

597

Forks

32

Language

Python

License

MIT

Last pushed

Aug 12, 2025

Commits (30d)

0

Get this data via API

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

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