Update app.py

app.py

@@ -309,94 +309,108 @@ def evaluate_predictions(prediction_file, model_name, add_to_leaderboard):
 
 initialize_leaderboard_file()
 
-with gr.Blocks() as demo:
+
+# Function to set default mode
+css_dark_mode = """
+body {
+    background-color: #1e1e1e;
+    color: #ffffff;
+}
+
+a {
+    color: #4caf50;
+}
+
+a:hover {
+    color: #81c784;
+    text-decoration: underline;
+}
+
+button {
+    background-color: #4caf50;
+    color: #ffffff;
+    border-radius: 5px;
+    padding: 10px;
+}
+
+button:hover {
+    background-color: #81c784;
+}
+
+.input-row, .tab-content {
+    background-color: #333333;
+    border-radius: 8px;
+    padding: 15px;
+}
+
+.dataframe {
+    color: #ffffff;
+    background-color: #2e2e2e;
+    border: 1px solid #4caf50;
+}
+"""
+
+with gr.Blocks(css=css_dark_mode) as demo:
     gr.Markdown("""
-# Competition Title
-### Welcome to the Competition Overview
+# 🏆 **Competition Title**
+### 🌟 **Welcome to the Competition Overview**
 ![Competition Logo](mobile_mmlu_sd.jpeg)
- Here you can submit your predictions, view the leaderboard, and track your performance!
+---
+Welcome to the **Mobile-MMLU Benchmark Competition**. Here you can submit your predictions, view the leaderboard, and track your performance!
+---
     """)
 
     with gr.Tabs():
         with gr.TabItem("📖 Overview"):
             gr.Markdown("""
-## Overview
-# Welcome to the Mobile-MMLU Benchmark Competition
-
-Evaluate the performance of mobile-compatible Large Language Models (LLMs) on 16,186 scenario-based and factual questions across 80 fields. Compete to showcase your model’s accuracy for real-world mobile scenarios.
-
-## What is Mobile-MMLU?
-
-Mobile-MMLU is a benchmark designed to test the capabilities of LLMs optimized for mobile use. By participating in this competition, you contribute to advancing mobile intelligence benchmarks and shaping the future of mobile-compatible AI systems.
-
+## 📘 Overview
+Welcome to the **Mobile-MMLU Benchmark Competition**! Evaluate mobile-compatible Large Language Models (LLMs) on **16,186 scenario-based and factual questions** across **80 fields**.
 ---
-
-## How It Works
-
-1. **Download the Dataset**  
-   Access the dataset and detailed generation instructions on our [GitHub page](https://github.com/your-github-repo).
-
-2. **Generate Predictions**  
-   Use your LLM to answer the questions and format your predictions as a CSV file with the following structure as written on our GitHub page :
-	 
-3. **Submit Predictions**  
-Upload your predictions via the submission portal.
-
-4. **Evaluation**  
-Your submission will be scored on accuracy 
-
-5. **Leaderboard**  
-Compare your results against other participants on the live leaderboard.
+### 🌐 **What is Mobile-MMLU?**
+Mobile-MMLU is a benchmark designed to test the capabilities of LLMs optimized for mobile use. Contribute to advancing mobile AI systems by competing to achieve the highest accuracy.
+
+### 🔍 **How It Works**
+1. **Download the Dataset**
+   Access the dataset and instructions on our [GitHub page](https://github.com/your-github-repo).
+2. **Generate Predictions**
+   Use your LLM to answer the dataset questions. Format your predictions as a CSV file.
+3. **Submit Predictions**
+   Upload your predictions on this platform.
+4. **Evaluation**
+   Submissions are scored on accuracy.
+5. **Leaderboard**
+   View real-time rankings on the leaderboard.
 
 ---
-
-## Competition Tasks
-
-Participants are tasked with generating predictions for the dataset and optimizing their models for:
-- **Accuracy**: Correctly answering questions across diverse fields.
+### 🏆 **Competition Tasks**
+Participants must:
+- Optimize their models for **accuracy**.
+- Answer diverse field questions effectively.
 ---
+### 🚀 **Get Started**
+1. Prepare your model using resources on our [GitHub page](https://github.com/your-github-repo).
+2. Submit predictions in the required format.
+3. Track your progress on the leaderboard.
 
-
-## Get Started
-
-1. **Prepare Your Model**  
-Refer to our [GitHub page](https://github.com/your-github-repo) for dataset access and response generation instructions.
-
-2. **Submit Predictions**  
-Format your submission as specified in the rules.
-
-3. **Track Progress**  
-Check the leaderboard for real-time rankings.
-
+### 📧 **Contact Us**
+For support, email: [Insert Email Address]
 ---
-
-## Contact Us
-
-For questions or support, contact us at: [Insert Email Address]
-                """)
+            """)
 
         with gr.TabItem("📤 Submission"):
             with gr.Row():
-                file_input = gr.File(label="Upload Prediction CSV", file_types=[".csv"], interactive=True)
-                model_name_input = gr.Textbox(label="Model Name", placeholder="Enter your model name")
+                file_input = gr.File(label="📂 Upload Prediction CSV", file_types=[".csv"], interactive=True)
+                model_name_input = gr.Textbox(label="🖋️ Model Name", placeholder="Enter your model name")
 
             with gr.Row():
-                overall_accuracy_display = gr.Number(label="Overall Accuracy", interactive=False)
-                add_to_leaderboard_checkbox = gr.Checkbox(label="Add to Leaderboard?", value=True)
+                overall_accuracy_display = gr.Number(label="🏅 Overall Accuracy", interactive=False)
+                add_to_leaderboard_checkbox = gr.Checkbox(label="📊 Add to Leaderboard?", value=True)
 
-            eval_button = gr.Button("Evaluate")
-            eval_status = gr.Textbox(label="Evaluation Status", interactive=False)
-
-            def handle_evaluation(file, model_name, add_to_leaderboard):
-                status, leaderboard = evaluate_predictions(file, model_name, add_to_leaderboard)
-                if leaderboard.empty:
-                    overall_accuracy = 0
-                else:
-                    overall_accuracy = leaderboard.iloc[-1]["Overall Accuracy"]
-                return status, overall_accuracy
+            eval_button = gr.Button("Evaluate", elem_id="evaluate-button")
+            eval_status = gr.Textbox(label="📢 Evaluation Status", interactive=False)
 
             eval_button.click(
-                handle_evaluation,
+                evaluate_predictions,
                 inputs=[file_input, model_name_input, add_to_leaderboard_checkbox],
                 outputs=[eval_status, overall_accuracy_display],
             )
@@ -404,18 +418,17 @@ For questions or support, contact us at: [Insert Email Address]
         with gr.TabItem("🏅 Leaderboard"):
             leaderboard_table = gr.Dataframe(
                 value=load_leaderboard(),
-                label="Leaderboard",
+                label="🏆 Leaderboard",
                 interactive=False,
                 wrap=True,
             )
-            refresh_button = gr.Button("Refresh Leaderboard")
+            refresh_button = gr.Button("🔄 Refresh Leaderboard")
             refresh_button.click(
                 lambda: load_leaderboard(),
                 inputs=[],
                 outputs=[leaderboard_table],
             )
 
-    gr.Markdown(f"Last updated on **{LAST_UPDATED}**")
+    gr.Markdown(f"**📅 Last updated:** {LAST_UPDATED}")
 
 demo.launch()
-