Update app.py

app.py

@@ -102,7 +102,7 @@
 
 
 
-
+# with bubble plot
 
 # import streamlit as st
 # import pandas as pd
@@ -224,8 +224,96 @@
 #     )
 
 
+# with all bubble plot
+
+
+# import streamlit as st
+# import pandas as pd
+# import altair as alt
+
+# # Set page layout to wide mode
+# st.set_page_config(
+#     page_title="PathVLMs Leaderboard",
+#     page_icon="🏆",
+#     layout="wide"
+# )
+
+# # Load the leaderboard data
+# df = pd.read_csv("leaderboard.csv")  # Replace with the actual filename
+
+# # Add title and description
+# st.title("PathVLMs Leaderboard 🏆")
+# st.markdown("""
+# Welcome to the **PathVLMs Leaderboard**! This leaderboard displays evaluation results for various Vision-Language Models (VLMs) in Pathology, focusing on multiple-choice questions (MCQs), answers, and explanations.  
+# You can search, filter, and visualize metrics for better insights.
+# """)
+
+# # Sidebar Filters
+# with st.sidebar:
+#     st.subheader("Filters")
+#     # Search by model name
+#     search_query = st.text_input("Search for Model Name", value="", placeholder="e.g., llava")
+    
+#     # Filter by model size
+#     model_sizes = df['Params (B)'].unique()
+#     selected_sizes = st.multiselect("Select Model Sizes", options=model_sizes, default=model_sizes)
+    
+#     # Filter by model type
+#     if 'Language Model' in df.columns:
+#         model_types = df['Language Model'].unique()
+#         selected_types = st.multiselect("Select Model Types", options=model_types, default=model_types)
+#     else:
+#         selected_types = []
+
+# # Apply Filters
+# filtered_df = df[
+#     (df['Params (B)'].isin(selected_sizes)) &
+#     (df['Language Model'].isin(selected_types) if selected_types else True) &
+#     (df['Method'].str.contains(search_query, case=False, na=False))
+# ]
+
+# # Main Leaderboard Section
+# st.subheader("Leaderboard Table")
+# if filtered_df.empty:
+#     st.warning("No results found. Try adjusting the filters.")
+# else:
+#     # Display the filtered table
+#     st.dataframe(filtered_df)
+
+#     # Dataset columns to plot
+#     dataset_columns = ['Socialpath Tiny', 'Socialpath All', 'Education Content Tiny', 'Education Content All', 'Pubmed Tiny', 'Pubmed All', 'Avg score']
+
+#     # Generate Bubble Plots for each dataset column
+#     for dataset in dataset_columns:
+#         st.subheader(f"Bubble Plot: {dataset}")
+#         bubble_chart = alt.Chart(filtered_df).mark_circle(size=200).encode(
+#             x=alt.X('Params (B):Q', title="Model Size (in Billion Params)"),
+#             y=alt.Y(f'{dataset}:Q', title=dataset),
+#             size=alt.Size('Params (B):Q', legend=None),
+#             color=alt.Color('Method:N', title="Model"),
+#             tooltip=['Method', 'Params (B)', dataset]
+#         ).properties(
+#             width=800,
+#             height=600
+#         )
+#         st.altair_chart(bubble_chart, use_container_width=True)
+
+#     # Download Button
+#     @st.cache
+#     def convert_df_to_csv(dataframe):
+#         return dataframe.to_csv(index=False).encode('utf-8')
+
+#     csv_data = convert_df_to_csv(filtered_df)
+#     st.download_button(
+#         label="Download Filtered Results",
+#         data=csv_data,
+#         file_name="filtered_leaderboard.csv",
+#         mime="text/csv"
+#     )
+
 
 
+# this also with bubble 
 
 import streamlit as st
 import pandas as pd
@@ -239,7 +327,7 @@ st.set_page_config(
 )
 
 # Load the leaderboard data
-df = pd.read_csv("leaderboard.csv")  # Replace with the actual filename
+df = pd.read_csv("leaderboard.csv")  # Replace with your actual CSV file name
 
 # Add title and description
 st.title("PathVLMs Leaderboard 🏆")
@@ -277,8 +365,28 @@ st.subheader("Leaderboard Table")
 if filtered_df.empty:
     st.warning("No results found. Try adjusting the filters.")
 else:
-    # Display the filtered table
-    st.dataframe(filtered_df)
+    # Display table in wide layout
+    st.dataframe(
+        filtered_df,
+        height=600,  # Adjust table height
+        width=1600  # Adjust table width
+    )
+
+    # Visualization of selected metric
+    st.subheader("Performance Metrics Visualization")
+    metric = st.selectbox("Select Metric to Visualize", options=filtered_df.columns[5:])
+
+    # Visualization Chart
+    chart = alt.Chart(filtered_df).mark_bar().encode(
+        x=alt.X('Method', sort=alt.EncodingSortField(field=metric, order='descending'), title="Model"),
+        y=alt.Y(metric, title=metric),
+        color='Method',
+        tooltip=['Method', metric]
+    ).properties(
+        width=1400,  # Full width
+        height=600   # Increased height
+    )
+    st.altair_chart(chart, use_container_width=True)
 
     # Dataset columns to plot
     dataset_columns = ['Socialpath Tiny', 'Socialpath All', 'Education Content Tiny', 'Education Content All', 'Pubmed Tiny', 'Pubmed All', 'Avg score']
@@ -313,3 +421,4 @@ else:
 
 
 
+