onnxruntime
Bring back SLN cuda kernel and use provider options to switch to standard implementation
#15660
Merged
Merged