fix: alpha to value

Author: kozistr

pytorch_optimizer/optimizer/adams.py

@@ -148,6 +148,6 @@ def step(self, closure: CLOSURE = None) -> LOSS:
                 de_nom = exp_avg_sq_hat.sqrt().add(group['eps'])
 
                 step_size = group['lr'] if self.adamd_debias_term else group['lr'] / bias_correction1
-                p.addcdiv_(exp_avg, de_nom, alpha=-step_size)
+                p.addcdiv_(exp_avg, de_nom, value=-step_size)
 
         return loss