#8686th of 105 in ML frameworks and libraries
autograd
Automatic differentiation for native Python and NumPy code
- Stars
- 7.5k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Jun 2026
- Language
- Python
7,449 on 1 Apr 20267,546 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Autograd differentiates Python and NumPy functions, including code with loops, ifs, recursion and closures. It supports reverse-mode and forward-mode differentiation, which can be composed to any derivative order. It is aimed at gradient-based optimization, with examples for neural nets, LSTMs, Gaussian processes and a fluid simulation.
Who it is for: Python and NumPy users needing gradients for optimization or research code
Strengths
- Differentiates plain Python and NumPy code, including loops, recursion and closures
- Reverse- and forward-mode can be composed for higher-order derivatives
- Installs with pip; SciPy support is an optional extra
- Examples cover CNN, RNN, LSTM, Bayesian neural net and Gaussian process
Weaknesses
- README does not mention GPU support; it is built around NumPy
- Reverse-mode is efficient only for scalar-valued functions of array arguments
- Requires importing the thinly-wrapped autograd.numpy instead of plain numpy
- No built-in neural network layers or optimizers described in the README
What it needs
- Needs numpy, scipy (optional)
Also in ML frameworks and libraries
See all 105| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 92 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 91 out of 100 |