ColorJitter #60
Description
Hello, I would like to ask which specific details in the pre-training process have a significant impact on the results? For example, corlorjitter? And other details as well?
Also, what are the respective benefits in Data Processing (Section 4.2.2)?
Thanks
At the same time, I would like to know whether the predicted actions and the true values are normalized or de-normalized when using L1 for evaluation?