[BUG] "Expected all tensors to be on the same device" when not specifying sample_weight and using cuda/gpu version of HMM

[tobemo]

Describe the bug
When fitting an uninitialized HMM, and not specifying sample_weight I get a runtime error saying that not all tensors are on the same device.
The error occurs on line 262 of distributions/normal.py

    261 if self.covariance_type == 'full':
--> 262 	self._w_sum += torch.sum(sample_weight, dim=0)
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

To Reproduce

Note that I'm not sure if this is the correct way of fitting an HMM using this package.

import pandas as pd
import numpy as np
import torch

from pomegranate.hmm import DenseHMM
from pomegranate.distributions import Normal

# taken from a hmmlearn example
# https://hmmlearn.readthedocs.io/en/latest/auto_examples/plot_poisson_hmm.html#sphx-glr-auto-examples-plot-poisson-hmm-py
X = pd.DataFrame(
    np.array([13, 14, 8, 10, 16, 26, 32, 27, 18, 32, 36, 24, 22, 23, 22, 18,
    25, 21, 21, 14, 8, 11, 14, 23, 18, 17, 19, 20, 22, 19, 13, 26,
    13, 14, 22, 24, 21, 22, 26, 21, 23, 24, 27, 41, 31, 27, 35, 26,
    28, 36, 39, 21, 17, 22, 17, 19, 15, 34, 10, 15, 22, 18, 15, 20,
    15, 22, 19, 16, 30, 27, 29, 23, 20, 16, 21, 21, 25, 16, 18, 15,
    18, 14, 10, 15, 8, 15, 6, 11, 8, 7, 18, 16, 13, 12, 13, 20,
    15, 16, 12, 18, 15, 16, 13, 15, 16, 11, 11])
)

X_ = torch.from_numpy(X.to_numpy())
X_ = X_.unsqueeze(0)  # add batch dimension
X_ = X_.cuda()
X_.shape
# torch.Size([1, 107, 1])

n_components = 3
model = DenseHMM([Normal()] * n_components, max_iter=100, verbose=True)
model = model.cuda()
assert model.device == X_.device

model.fit(X_)

When I use the below I get another error, for which I will open a separate issue.