Update app.py

app.py

@@ -6,26 +6,21 @@ from yaml import safe_load
 import pandas as pd
 import gradio as gr
 
-# 加载配置文件
 CONFIG = safe_load(open("config.yaml"))
 label_map = {'Avg':"All", "API":"Web API", "Code": "Code Function", "Customized": "Customized App"}
-# 读取数据并进行初步处理
+
 data = defaultdict(dict)
 for setting in CONFIG['settings']:
     for data_type in CONFIG['types']:
         file_path = os.path.join("data", f"{CONFIG['settings_mapping'][setting]}-{data_type}.xlsx")
         df = pd.read_excel(file_path)
         
-        # 添加平均分列，计算除第一列和倒数两列之外的均值
         df["Average"] = df.iloc[:, 1:-2].mean(axis=1)
         
-        # 添加 Rank 列，根据 Average 降序排名
         df["Rank"] = df["Average"].rank(ascending=False, method='min').astype(int)
         
-        # 按 Rank 排序（Rank 值越小越靠前）
         df = df.sort_values("Rank", ascending=True)
         
-        # 将列重新排序：第一列为 Rank，第二列为 Model，第三列为 Average，其余列保持原有顺序
         cols = df.columns.tolist()
         first_cols = []
         if "Rank" in cols:
@@ -37,7 +32,6 @@ for setting in CONFIG['settings']:
         remaining_cols = [col for col in cols if col not in first_cols]
         df = df[first_cols + remaining_cols]
         
-        # 数值格式化：对于数值列（除 Rank 列），如果最大值 <= 1 则认为是比例数据（乘以 100 后保留两位小数），否则直接保留两位小数
         numeric_cols = df.select_dtypes(include=['float', 'int']).columns
         for col in numeric_cols:
             if col != "Rank":
@@ -48,9 +42,7 @@ for setting in CONFIG['settings']:
                     
         data[setting][data_type] = df
 
-# 自定义 CSS 样式，包括表格样式及标签页的边框美化
 css = """
-/* 表格样式 */
 table thead th, table thead td {
     text-align: center !important;
 }
@@ -74,7 +66,6 @@ table > tbody > tr > td:not(:nth-child(2)) {
     vertical-align: middle;
 }
 
-/* 外层标签页增加边框、内边距和圆角 */
 .outer-tabs {
     border: 2px solid #ccc;
     border-radius: 8px;
@@ -94,7 +85,6 @@ table > tbody > tr > td:not(:nth-child(2)) {
     border-bottom: 2px solid #0078d7;
 }
 
-/* 内层标签页增加边框、内边距和圆角 */
 .inner-tabs {
     border: 2px solid #aaa;
     border-radius: 8px;
@@ -115,7 +105,6 @@ table > tbody > tr > td:not(:nth-child(2)) {
 }
 """
 
-# 模型类型和模型大小（数值区间）设置
 MODEL_TYPES = [
     "sparse retrieval",
     "dense retrieval",
@@ -131,24 +120,21 @@ NUMERIC_INTERVALS = {
     ">1B": pd.Interval(1000, 1_000_000, closed='right'),
 }
 
-# 定义过滤函数，实现搜索、模型类型及模型大小过滤功能，并重新计算局部 Rank
+
 def filter_data(search_query, model_types, model_sizes):
     outputs = []
     for setting in CONFIG['settings']:
         for data_type in CONFIG['types']:
             df = data[setting][data_type].copy()
             
-            # 搜索过滤：在 "Model" 列中查找包含任一搜索关键字的记录
             if search_query:
                 queries = [q.strip().lower() for q in search_query.split(";") if q.strip()]
                 mask_search = df["Model"].str.lower().apply(lambda x: any(q in x for q in queries))
                 df = df[mask_search]
             
-            # 模型类型过滤：假设 Excel 中存在 "Model Type" 列
             if model_types and set(model_types) != set(MODEL_TYPES):
                 df = df[df["Model Type"].isin(model_types)]
             
-            # 模型大小过滤：将 "Number of Parameters" 转换为数值，并利用选定的区间进行过滤
             def parse_params(val):
                 try:
                     if isinstance(val, str):
@@ -176,11 +162,9 @@ def filter_data(search_query, model_types, model_sizes):
             if "params_numeric" in df.columns:
                 df = df.drop(columns=["params_numeric"])
             
-            # 重新计算 Rank，根据当前过滤后的 Average 进行排序（局部 Rank）
             df["Rank"] = df["Average"].rank(ascending=False, method='min').astype(int)
             df = df.sort_values("Rank", ascending=True)
             
-            # 重新排列列顺序：Rank, Model, Average, 其他
             cols = df.columns.tolist()
             first_cols = []
             if "Rank" in cols:
@@ -195,7 +179,7 @@ def filter_data(search_query, model_types, model_sizes):
             outputs.append(df)
     return outputs
 
-# 创建 Gradio 界面
+
 with gr.Blocks(css=css) as demo:
     gr.Markdown("""
         ## Tool-Retrieval benchmark leaderboard 
@@ -233,7 +217,6 @@ with gr.Blocks(css=css) as demo:
     
     submit_button = gr.Button("Filter Data")
     
-    # 创建嵌套标签页，外层标签页使用 outer-tabs 类，内层标签页使用 inner-tabs 类
     output_dfs = []
     with gr.Tabs(elem_classes="outer-tabs") as result_tabs:
         for setting in CONFIG['settings']:
@@ -244,7 +227,6 @@ with gr.Blocks(css=css) as demo:
                             df_component = gr.DataFrame(value=data[setting][data_type], type="pandas")
                             output_dfs.append(df_component)
     
-    # 将过滤函数与按钮绑定，点击后更新所有 DataFrame 组件
     submit_button.click(
          fn=filter_data,
          inputs=[search_box, model_type_checkbox_group, model_size_checkbox_group],