update

app.py

@@ -11,6 +11,15 @@ def case_insensitive_search(data, query, column):
         return data[data[column].str.lower().str.contains(query.lower())]
     return data
 
+# 创建一个带有滚动条的表格显示函数
+def display_table(data, rows_per_page=20):
+    # 使用 Streamlit 的 container 来创建滚动效果
+    container = st.container()
+    with container:
+        # 设置表格高度，基于每行的大致高度估算
+        height = min(40 + rows_per_page * 38, 800)  # 最大高度为800像素
+        st.dataframe(data, height=height)
+
 # 页面布局和功能
 def main():
     st.title("Multihop-RAG Benchmark Space")
@@ -19,24 +28,27 @@ def main():
 
     # 添加搜索框
     st.sidebar.header("Search Options")
-    framework_query = st.sidebar.text_input("Search by Framework")
-    model_query = st.sidebar.text_input("Search by Model")
+    chat_model_query = st.sidebar.text_input("Search by Chat Model")
+    embedding_model_query = st.sidebar.text_input("Search by Embedding Model")
+    chunk_query = st.sidebar.text_input("Search by Chunk")  # 新增按 Chunk 搜索
 
     # 根据输入执行搜索
-    if framework_query:
-        data = case_insensitive_search(data, framework_query, 'framework')
-    if model_query:
-        data = case_insensitive_search(data, model_query, 'model')
+    if chat_model_query:
+        data = case_insensitive_search(data, chat_model_query, 'chat_model')
+    if embedding_model_query:
+        data = case_insensitive_search(data, embedding_model_query, 'embedding_model')
+    if chunk_query:  # 新增 Chunk 的筛选
+        data = case_insensitive_search(data, chunk_query, 'chunk')
 
     # 显示数据
     st.header("Benchmark Results")
-    st.write("Displaying results for MRR@10 and Hit@10 across different frameworks, models, and chunk sizes.")
-    st.dataframe(data)
+    st.write("Displaying results for MRR@10, Hit@10, and Accuracy across different frameworks, embedding models, chat models, and chunks.")
+    display_table(data)  # 使用自定义的表格显示函数
 
     # 数据统计和图表
     if st.sidebar.checkbox("Show Metrics Distribution"):
         st.subheader("Metrics Distribution")
-        st.bar_chart(data[['MRR@10', 'Hit@10']])
+        st.bar_chart(data[['MRR@10', 'Hit@10', 'Accuracy']])  # 添加Accuracy到图表中
 
 if __name__ == "__main__":
     main()

benchmark_data.csv

@@ -1,11 +1,9 @@
-framework,model,chunk size,MRR@10,Hit@10
-Transformers,BERT-base,128,0.32,0.45
-Transformers,BERT-large,256,0.37,0.50
-Fairseq,ROBERTA-base,128,0.35,0.48
-Fairseq,ROBERTA-large,256,0.40,0.55
-HuggingFace,GPT-2,128,0.30,0.44
-HuggingFace,GPT-3,256,0.42,0.57
-TensorFlow,T5-small,128,0.31,0.46
-TensorFlow,T5-large,256,0.39,0.51
-PyTorch,BART-base,128,0.33,0.47
-PyTorch,BART-large,256,0.38,0.53
+framework,chat_model,embedding_model,chunk,MRR@10,Hit@10,Accuracy
+Framework1,ChatModelA,EmbeddingModel1,Chunk1,0.32,0.78,0.88
+Framework2,ChatModelB,EmbeddingModel2,Chunk2,0.25,0.65,0.90
+Framework1,ChatModelC,EmbeddingModel1,Chunk1,0.37,0.85,0.92
+Framework3,ChatModelA,EmbeddingModel3,Chunk3,0.29,0.75,0.89
+Framework2,ChatModelB,EmbeddingModel2,Chunk2,0.31,0.70,0.87
+Framework3,ChatModelC,EmbeddingModel3,Chunk3,0.35,0.80,0.93
+Framework1,ChatModelA,EmbeddingModel1,Chunk1,0.33,0.82,0.91
+Framework2,ChatModelB,EmbeddingModel2,Chunk2,0.26,0.68,0.85