cvxpy/cvxpylayers

Health v2 · maintenance only

Adoption not part of the score

2126 stars · 193 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

CVXPYlayers is a Python library for constructing differentiable convex optimization layers in PyTorch, JAX, and MLX using CVXPY. It solves parametrized convex problems in the forward pass and computes derivatives of the solution with respect to parameters in the backward pass, with GPU acceleration via Moreau and CuClarabel backends.

Use cases

• embed a convex optimization problem as a layer in a neural network
• backpropagate through a parametrized convex program in PyTorch or JAX
• solve convex optimization problems on the GPU inside a training loop
• learn parameters of a quadratic program or cone program end-to-end
• build model-predictive control policies that are differentiable
• train models whose forward pass includes a constrained optimization step

When to choose

• your model includes a convex optimization step and you need gradients through it
• you want to express problems in CVXPY and use them as differentiable layers
• you need GPU-accelerated convex solving within deep learning frameworks

When to avoid

• your optimization problem is nonconvex
• you only need to solve convex problems without gradients (plain CVXPY suffices)
• you need discrete or combinatorial optimization layers

Facets

library · maturity active

machine-learning math sdk machine-learning deep-learning gpu-computing python convex-optimization differentiable-optimization cvxpy pytorch jax mlx optimization-layers gpu-acceleration algorithms gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="cvxpy/cvxpylayers")