PythonOT/POT
Python library providing differentiable optimal transport solvers for machine learning applications including domain adaptation, GNN layers, and OT barycenters.
Velocity · 7d
+0.8
★ / day
Trend
→steady
star history
POT provides a comprehensive set of solvers for optimal transport problems including exact linear OT, regularized OT (Sinkhorn), Gromov-Wasserstein, and fused variants. It offers ML-specific tools for domain adaptation, optimal transport mapping estimation, and GNN layers. The library supports multiple backends (PyTorch, Jax, TensorFlow, NumPy, CuPy) enabling differentiable computations for integration into ML pipelines.