Wrong state inference?

[hunglv51]

Hello TreB1eN,

I have a question about the code that predict the hidden state in viterbi_inference

HiddenMarkovModel_Pytorch/HiddenMarkovModel.py

Lines 66 to 68 in 50aed18

	state = states_seq[step]
	state_prob = pathStates[step][state]
	states_seq[step -1] = state_prob

I think to predict state, we have to choose the state with the highest probability like code that predict last state

HiddenMarkovModel_Pytorch/HiddenMarkovModel.py

Line 63 in 50aed18

states_seq[self.N - 1] = torch.argmax(pathScores[self.N-1, :], 0)

Can you explain it to me?

Thanks!