Open
Description
Over the past weeks, several IBs failed while building Rivet. The issue was caused by instability of the website hosting the ONNX model (currently only one, though the Makefile is set up to fetch all available files). CERN IT has addressed the immediate problem, but to prevent similar issues in the future, we would like to clarify:
(a) Do we actually want to build the ONNX analyses in our builds?
(b) If yes, can/should we mirror these models on cmssdt to ensure stability?