Update app.py

app.py

@@ -11,14 +11,6 @@ data = {
         "LLaVa-v1.6-Mistral (7B)",
         "Gemini 1.5 Flash (MLLM)", "Gemini 2.0 Flash (MLLM)", "Gemini 1.5 Pro (MLLM)", "Gemini 2.0 Pro (MLLM)", "GPT-4o (MLLM)"
     ],
-    "Type": [
-        "LLM", "LLM", "LLM", "LLM", "LLM",
-        "LLM", "LLM", "LLM", "LLM",
-        "LLM", "LLM", "LLM", "LLM",
-        "LLM", "LLM", "LLM",
-        "MLLM",
-        "MLLM", "MLLM", "MLLM", "MLLM", "MLLM"
-    ],
     "T (Full)": [63.89, 64.91, 57.80, 63.37, 62.83, 49.94, 57.60, 56.06, 65.65, 56.94, 65.32, 56.90, 58.36, 62.78, 60.75, 57.43, None, 64.91, 64.79, 66.22, 66.42, 64.95],
     "TA (Full)": [45.81, 47.10, 40.69, 45.51, 45.82, 34.78, 39.86, 41.01, 47.42, 40.66, 43.38, 38.20, 37.47, 43.25, 44.59, 37.64, None, 45.06, 40.07, 46.90, 46.17, 44.53],
     "TAC (Full)": [21.29, 23.65, 19.08, 19.29, 22.67, 14.13, 16.49, 19.54, 19.65, 19.75, 22.54, 18.97, 19.20, 21.26, 18.63, 15.35, None, 20.66, 19.74, 23.23, 18.25, 19.60],
@@ -27,70 +19,16 @@ data = {
     "TAC (Segmented)": [21.36, 28.17, 21.81, 23.96, 25.38, 19.83, 19.56, 20.76, 22.77, 21.54, 24.60, 21.31, 21.43, 22.87, 24.20, 24.50, 3.30, 23.27, 23.52, 24.97, 25.21, 27.86]
 }
 
+# Create DataFrame
 df = pd.DataFrame(data)
 
-def filter_leaderboard(view="Segmented", model_type="All"):
-    columns = ["Model", "Type"]
-    if view == "Segmented":
-        columns += ["T (Segmented)", "TA (Segmented)", "TAC (Segmented)"]
-    else:
-        columns += ["T (Full)", "TA (Full)", "TAC (Full)"]
-    
-    if model_type == "All":
-        return df[columns].sort_values(by=columns[-1], ascending=False)
-    else:
-        return df[df["Type"] == model_type][columns].sort_values(by=columns[-1], ascending=False)
+def display_leaderboard():
+    return df
 
+# Create a simple Gradio interface
 with gr.Blocks(title="VideoConviction LLM Leaderboard") as demo:
-    gr.Markdown("# 📊 VideoConviction Benchmark Leaderboard")
-    gr.Markdown("""
-This leaderboard displays the performance (F1 scores) of various Large Language Models (LLMs) and Multimodal LLMs (MLLMs) on the **VideoConviction** benchmark.
-
----
-
-### 📘 About the Benchmark
-**VideoConviction** is a multimodal benchmark designed to evaluate **human conviction** and **stock market recommendation quality** from video content. It integrates vision, audio, and text to evaluate reasoning and persuasion detection in financial decision-making.
-
-📝 **Paper**: [VideoConviction: A Multimodal Benchmark for Human Conviction and Stock Market Recommendations](https://doi.org/10.1145/3711896.3737417)  
-📍 **Conference**: ACM SIGKDD 2025  
-👥 **Authors**: Michael Galarnyk, Veer Kejriwal, Agam Shah, Yash Bhardwaj, Nicholas Watney Meyer, Anand Krishnan, Sudheer Chava
-
----
-
-### 📄 Citation
-```bibtex
-@inproceedings{galarnyk2025videoconviction,
-  author = {Michael Galarnyk and Veer Kejriwal and Agam Shah and Yash Bhardwaj and Nicholas Watney Meyer and Anand Krishnan and Sudheer Chava},
-  title = {VideoConviction: A Multimodal Benchmark for Human Conviction and Stock Market Recommendations},
-  booktitle = {Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD '25)},
-  year = {2025},
-  location = {Toronto, ON, Canada},
-  pages = {12},
-  publisher = {ACM},
-  doi = {10.1145/3711896.3737417}
-}
-```""")
-
-with gr.Row():
-    with gr.Column():
-        view_dropdown = gr.Dropdown(
-            label="Select View Type", choices=["Segmented", "Full"], value="Segmented"
-        )
-    with gr.Column():
-        model_dropdown = gr.Dropdown(
-            label="Filter by Model Type", choices=["All", "LLM", "MLLM"], value="All"
-        )
-
-leaderboard = gr.Dataframe(
-    value=filter_leaderboard("Segmented", "All"),
-    label="Leaderboard",
-    interactive=False
-)
-
-def update_leaderboard(view, model_type):
-    return filter_leaderboard(view, model_type)
-
-view_dropdown.change(fn=update_leaderboard, inputs=[view_dropdown, model_dropdown], outputs=leaderboard)
-model_dropdown.change(fn=update_leaderboard, inputs=[view_dropdown, model_dropdown], outputs=leaderboard)
+    gr.Markdown("# VideoConviction Benchmark Leaderboard")
+    gr.Markdown("This leaderboard shows the F1 scores of various LLMs and MLLMs across the VideoConviction benchmark tasks.")
+    gr.Dataframe(display_leaderboard, interactive=False)
 
 demo.launch()