Error during nnRNN initialization

[AbdelkaderMH]

Traceback (most recent call last):
File "sMNIST.py", line 281, in
rnn = select_network(args, inp_size)
File "/content/nnRNN/utils.py", line 82, in select_network
i_initializer=iinit)
File "/content/nnRNN/RNN_Cell.py", line 143, in init
self.reset_parameters()
File "/content/nnRNN/RNN_Cell.py", line 161, in reset_parameters
torch.as_tensor(self.r_initializer(self.hidden_size), dtype=torch.float32))
File "/content/nnRNN/RNN_Cell.py", line 173, in _B
return expm(self._A())
File "/content/nnRNN/expRNN/exp_numpy.py", line 533, in expm
return _expm(A, use_exact_onenorm=True)
File "/content/nnRNN/expRNN/exp_numpy.py", line 590, in _expm
s = max(int(torch.ceil(torch.log2(eta_5 / theta_13))), 0)
ValueError: cannot convert float NaN to integer