PyPI page
Home page
Author:
None
Summary:
**thoad** (Torch High Order Automatic Differentiation) is a lightweight reverse-mode autodifferentiation engine written entirely in Python that works over PyTorch’s computational graph to compute **high order partial derivatives**. Unlike PyTorch’s native autograd - which is limited to first-order native partial derivatives - **thoad** is able to performantly propagate arbitray-order derivatives throughout the graph, enabling more advanced gradient-based computations.
Latest version:
0.1.0
Required dependencies:
torch
Downloads last day:
2
Downloads last week:
6
Downloads last month:
11