diff --git a/readme.md b/readme.md
index c34791e..201c52c 100644
--- a/readme.md
+++ b/readme.md
@@ -13,11 +13,13 @@ https://gitea.pinblog.codes/attachments/9b2424f7-7e7d-4ad1-a368-86a523d67504
 * 원본
 https://gitea.pinblog.codes/attachments/9a31bb80-bc0a-4d5a-83b1-4ef0557456ad
 
+
 -----
 
 ### Dataset
 [Kaggle - WDI Data](https://www.kaggle.com/qingyi/wm811k-wafer-map/code)
 
+
 -----
 
 ### 수행방법
@@ -29,8 +31,8 @@ https://gitea.pinblog.codes/attachments/9a31bb80-bc0a-4d5a-83b1-4ef0557456ad
 
   https://gitea.pinblog.codes/CBNU/03_WDI_CNN/releases/tag/info
 
-  # Model
 
+# Model
 
 ```python
 import torch
@@ -99,7 +101,6 @@ cnn_wdi = CNN_WDI(class_num=9)
 
 # Load Data
 
-
 ```python
 from torchvision import transforms, datasets
 
@@ -124,7 +125,6 @@ test_dataset = datasets.ImageFolder(root='E:/wm_images/test/', transform=data_tr
 
 # Settings
 
-
 ```python
 import torch.optim as optim
 
@@ -149,11 +149,8 @@ val_max_images = 25
 
 ```
 
-    
-
 # Train Function
 
-
 ```python
 # 학습 함수 정의
 def train(model, dataloader, criterion, optimizer, device):
@@ -185,7 +182,6 @@ def train(model, dataloader, criterion, optimizer, device):
 
 # Evaluate Function
 
-
 ```python
 # 평가 함수 정의
 def evaluate(model, dataloader, criterion, device):
@@ -246,14 +242,15 @@ for epoch in range(num_epochs + 1):
         torch.save(cnn_wdi.state_dict(), 'CNN_WDI_' + str(epoch) + 'epoch.pth')
 ```
 
+
 -----
 
 ### 평가방법
 
 * 모델의 성능지표(Precision, Recall, Accuracy, F1-Score)를 혼동행렬(Confusion Metrix)로 구현한다.
 
-# Confusion Metrix
 
+# Confusion Metrix
 
 ```python
 import numpy as np
@@ -340,7 +337,6 @@ def predict_and_plot_metrics(title, model, dataloader, criterion, device):
 
 # Evaluate
 
-
 ```python
 import os
 import re
@@ -368,7 +364,6 @@ for model in sorted_models:
 
 # Loss Graph
 
-
 ```python
 import matplotlib.pyplot as plt
 
@@ -412,7 +407,6 @@ plt.show()
 
 # Print Selecting Test Model Result
 
-
 ```python
 def output(model, dataloader, criterion, device):
     model.eval()