GP-VAE met error when running on CUDA

[WenjieDu]

1. System Info

pypots==0.1.3

2. Information

• The official example scripts
• My own created scripts

3. Reproduction

Run a instantiated GP-VAE on CUDA device(s).

4. Expected behavior

PyPOTS/pypots/imputation/gpvae/model.py

Lines 175 to 176 in 66ace35

	def kl_divergence(a, b):
	return torch.distributions.kl.kl_divergence(a, b)

a and b are not on the same device. a is on CUDA and b is on CPU.