diff --git a/code/conversion/setup.conf b/code/conversion/setup.conf
index 0361197..d152668 100644
--- a/code/conversion/setup.conf
+++ b/code/conversion/setup.conf
@@ -4,23 +4,23 @@ train{
     fileprefix_list = [
 broken_bullet_50k, # more input models can be added here
 ]
-    d_in = 3
-    plot_frequency = 5000
+    d_in = 3 # 输入数据的维度。在3D点云数据中，通常为3（x、y、z坐标）
+    plot_frequency = 5000 # 每5000次迭代绘制一次点云
     checkpoint_frequency = 5000
     status_frequency = 100
     weight_decay = 0
     learning_rate_schedule = [{
-                                "Type" : "Step",
+                                "Type" : "Step", # 学习率调度类型。"Step"表示在指定迭代次数后将学习率乘以因子
                                 "Initial" : 0.005,
                                 "Interval" : 2000,
                                 "Factor" : 0.5
                                 }]
-    network_class = model.network.NHRepNet
+    network_class = model.network.NHRepNet # 网络类型。NHRepNet是neural halfspace representation network的缩写
 }
 
 plot{
-    resolution = 128
-    mc_value = 0.0
+    resolution = 128 # 体素网格的分辨率。128表示体素网格的每个维度有128个单元格
+    mc_value = 0.0 # 体素网格的体素值。0.0表示体素网格的体素值为0
     is_uniform_grid = True
     verbose = False
     save_html = False
@@ -29,20 +29,20 @@ plot{
 }
 network{
     inputs{
-        dims_sdf = [256, 256, 256]
-        skip_in = []
-        geometric_init= True
-        radius_init = 1
-        beta=100
+        dims_sdf = [256, 256, 256] # 体素网格的维度。[256, 256, 256]表示体素网格的每个维度有256个单元格
+        skip_in = [] # 跳过输入的索引。[]表示不跳过任何输入
+        geometric_init= True # 几何初始化。True表示使用几何初始化
+        radius_init = 1 # 半径初始化。1表示半径初始化为1
+        beta=100 # beta值。100表示beta值为100
     }
     sampler{
-        sampler_type = NormalPerPoint
+        sampler_type = NormalPerPoint # 采样器类型。NormalPerPoint表示每个点都使用正态分布采样
         properties{
-            global_sigma = 1.8
+            global_sigma = 1.8 # 全局sigma值。1.8表示全局sigma值为1.8
             }
         }
     loss{
-        lambda = 1
-        normals_lambda = 1
+        lambda = 1 # 损失函数中的lambda值。1表示lambda值为1
+        normals_lambda = 1 # 损失函数中的normals_lambda值。1表示normals_lambda值为1
     }
 }