centroids = init_centroids.clone()
states = []
for s_idx in range(n_states):
precisions = torch.ones(2)
states.append(DiagNormalModel(centroids[s_idx], precisions))
hmm = HiddenMarkovModel(states, T0=T0, T=T)
converged = hmm.fit(X, epsilon=1e-8)
print()
print('CONVERGED', converged)
print()
score = hmm.log_prob(X) # + hmm.log_parameters_prob()
print('ll', score)
# print('ll (no prior)', hmm.log_prob(X))
print("Pi Matrix: ")
print(hmm.T0)
# print("Transition Matrix: ")
# print(hmm.T)
# assert np.allclose(transition.exp().data.numpy(), True_T, atol=0.1)
# print()
# print("Emission Matrix: ")
# for s in hmm.states:
# print("Means")
# print(list(s.parameters())[0])
