 Inquiry About Data Processing

Hello, I'm quite interested in your work. I noticed that in your code for pose output, the range of the translation vectors is limited to -1 to 1 due to the use of the tanh activation function. I'm considering applying this code to my dataset, which has a larger span of translation vectors. Could you please share how you processed your data to fit within this range?