Possible discrepancy in REWOBetaScheduler Formula (Lines 139-142)

[porink0424]

I noticed that the formula updating self.communication_loss_ema in REWOBetaScheduler (emecom_gen/model/game/beta_scheduler.py, lines 139-142) appears to differ from the REWO algorithm described in the original paper (https://arxiv.org/abs/1905.04982).
It seems that the coefficients might be applied in different terms.
If you have time, it might be worth checking. Thank you!