Hi, I saw the current best/current best window average metrics in the log, so what does the window metrics mean and which metrics did you use in your paper?
We only used window average as a more stable metric for us to track performance when experimenting with new ideas. We do not use window scores in the paper.
Hi, I saw the current best/current best window average metrics in the log, so what does the window metrics mean and which metrics did you use in your paper?