statusfailed/catgrad
a categorical deep learning compiler
This project helps machine learning engineers and researchers optimize their deep learning model training workflows. It takes a model definition and 'compiles' its reverse pass into highly efficient, static code. The output is a standalone training loop that no longer depends on a deep learning framework, leading to faster and more resource-efficient model iteration.
208 stars. No commits in the last 6 months. Available on PyPI.
Use this if you are a deep learning engineer or researcher looking to improve the performance and reduce the overhead of your model training processes by eliminating framework dependencies.
Not ideal if you prefer the convenience of dynamic autodifferentiation within existing deep learning frameworks and do not require the extreme optimization benefits of static compilation.
Stars
208
Forks
7
Language
Python
License
MIT
Category
Last pushed
Sep 29, 2025
Commits (30d)
0
Dependencies
2
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/statusfailed/catgrad"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
JonathanRaiman/theano_lstm
:microscope: Nano size Theano LSTM module
google/tangent
Source-to-Source Debuggable Derivatives in Pure Python
ahrefs/ocannl
OCANNL: OCaml Compiles Algorithms for Neural Networks Learning
yoshoku/numo-openblas
Numo::OpenBLAS builds and uses OpenBLAS as a background library for Numo::Linalg
pranftw/neograd
A deep learning framework created from scratch with Python and NumPy