Hi, I have implementation of weighted std
in NumPy
as:
def weighted_std(average_of_tensor_list, list_of_tensors, list_num_samples):
list_of_tensors_new = np.array([(list_of_tensors[i] - average_of_tensor_list)**2 for i in range(len(list_of_tensors))], dtype="object")
variance = np.average(list_of_tensors_new, weights=list_num_samples, axis=0)
std = np.array([np.sqrt(variance[i]) for i in range(len(variance))], dtype="object")
return std
I am trying to convert this into equivalent PyTorch
code, can anyone please guide, what should be the way?