diff --git a/train.py b/train.py
index 906681199863f23d76148032412092d87bf301e1..fecd21b126b334f90c4e1f73d2c31aa9b77c64d2 100644
--- a/train.py
+++ b/train.py
@@ -238,11 +238,15 @@ if __name__ == "__main__":
         Init_Epoch = 0
         Freeze_Epoch = 50
         
-        optimizer = optim.Adam(net.parameters(),lr,weight_decay=5e-4)
+        #----------------------------------------------------------------------------#
+        #   我在实际测试时，发现optimizer的weight_decay起到了反作用，
+        #   所以去除掉了weight_decay，大家也可以开起来试试，一般是weight_decay=5e-4
+        #----------------------------------------------------------------------------#
+        optimizer = optim.Adam(net.parameters(),lr)
         if Cosine_lr:
             lr_scheduler = optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=5, eta_min=1e-5)
         else:
-            lr_scheduler = optim.lr_scheduler.StepLR(optimizer,step_size=1,gamma=0.95)
+            lr_scheduler = optim.lr_scheduler.StepLR(optimizer,step_size=1,gamma=0.92)
 
         if Use_Data_Loader:
             train_dataset = YoloDataset(lines[:num_train], (input_shape[0], input_shape[1]), mosaic=mosaic, is_train=True)
@@ -275,11 +279,15 @@ if __name__ == "__main__":
         Freeze_Epoch = 50
         Unfreeze_Epoch = 100
 
-        optimizer = optim.Adam(net.parameters(),lr,weight_decay=5e-4)
+        #----------------------------------------------------------------------------#
+        #   我在实际测试时，发现optimizer的weight_decay起到了反作用，
+        #   所以去除掉了weight_decay，大家也可以开起来试试，一般是weight_decay=5e-4
+        #----------------------------------------------------------------------------#
+        optimizer = optim.Adam(net.parameters(),lr)
         if Cosine_lr:
             lr_scheduler = optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=5, eta_min=1e-5)
         else:
-            lr_scheduler = optim.lr_scheduler.StepLR(optimizer,step_size=1,gamma=0.95)
+            lr_scheduler = optim.lr_scheduler.StepLR(optimizer,step_size=1,gamma=0.92)
 
         if Use_Data_Loader:
             train_dataset = YoloDataset(lines[:num_train], (input_shape[0], input_shape[1]), mosaic=mosaic, is_train=True)