Possibly incorrect order of operations in normalization

[seann999]

I think it should be (x - mean) * std?

octo/octo/utils/gym_wrappers.py

Lines 283 to 285 in 37951e4

	def normalize(self, data, metadata):
	if self.normalization_type == "normal":
	return (data / (metadata["std"] + 1e-8)) - metadata["mean"]

[kpertsch]

Ah good catch! Fixed it in #15 .
This should only affect scenarios in which you pass proprio to the policy (which needs to get normalized), eg the ALOHA finetune example.