nyakslamo/yield-curve-intelligence

Yield Curve Intelligence Platform

Interactive U.S. Treasury yield curve analysis: Lagrange/linear/cubic-spline/PCHIP interpolation compared honestly (including where Lagrange catastrophically diverges on the real grid), leave-one-maturity-out validation, PCA factor analysis (level/slope/curvature, verified from real data), and anomaly detection against ~10 years of history.

Full source, tests, and documentation: https://github.com/lamsofttech/yield-curve-intelligence

This Space's requirements.txt adds torch/spaces to the base app's dependencies (not present in the GitHub repo's own requirements.txt, which stays lightweight for Docker/CI/local dev) -- this account's free Hugging Face tier requires ZeroGPU hardware for Gradio Spaces, and ZeroGPU requires at least one genuine @spaces.GPU-decorated function to start. See the app's "GPU Historical Stress Test" tab and app.py's module docstring for what that function actually does (a genuine, measured computation, not a no-op).

Educational/research use only -- see the Responsible Use tab in the app itself.