I'm trying the example of GaussianKernel-
mat = torch.randn(3, 3, dtype=torch.float64)
sym_mat = mat @ mat.T
K = falkon.kernels.GaussianKernel(sigma=sym_mat)
But I got the error-
''.../falkon/kernels/distance_kernel.py", line 33, in validate_sigma
raise ValueError("sigma must be a scalar or a vector.")
ValueError: sigma must be a scalar or a vector.
Is the current version of falkon accept matrix for the sigma argument? Could you help with this issue? FYI, I'm using falkon 0.8.4 with torch version 1.13.0 and python 3.9. There's no problem when I use a scalar or a vector for the argument sigma.
Hi,
I'm trying the example of GaussianKernel- mat = torch.randn(3, 3, dtype=torch.float64) sym_mat = mat @ mat.T K = falkon.kernels.GaussianKernel(sigma=sym_mat)
But I got the error- ''.../falkon/kernels/distance_kernel.py", line 33, in validate_sigma raise ValueError("sigma must be a scalar or a vector.") ValueError: sigma must be a scalar or a vector.
Is the current version of falkon accept matrix for the sigma argument? Could you help with this issue? FYI, I'm using falkon 0.8.4 with torch version 1.13.0 and python 3.9. There's no problem when I use a scalar or a vector for the argument sigma.
Could you help with this? Thanks!