mmlu_data_to_example.py

import gradio as gr
import requests
import pandas as pd

class MultiURLData:
    def __init__(self, file_path):
        self.file_path = file_path

    def fetch_data(self):
        # Read URLs from a file, one per line
        with open(self.file_path, 'r') as f:
            file_urls = [line.strip() for line in f.readlines()]

        dataframes = []
        for url in file_urls:
            # Derive column names from the URLs
            column_name = url.split('/')[-1].split('_')[0]

            # Load data from URL
            response = requests.get(url)
            data = response.json()

            # Convert data into a DataFrame
            df = pd.DataFrame(data['results']).T

            # Rename 'acc' column to respective file names
            df = df.rename(columns={'acc': column_name})

            # Remove 'hendrycksTest-' from the index
            df.index = df.index.str.replace('hendrycksTest-', '')

            dataframes.append(df[[column_name]])  # keep only the column of interest

        # Merge the dataframes
        # Merge the dataframes on index
        data = pd.concat(dataframes, axis=1)

        # Transpose the dataframe to swap rows and columns
        data = data.transpose()

        # Select only columns 'moral_scenarios' and 'moral_disputes'
        data = data[['moral_scenarios', 'moral_disputes']]

        return data

data_provider = MultiURLData("file_urls.txt")

block = gr.Blocks()

with block:    
    gr.Markdown("""Leaderboard""")
    with gr.Tabs():
        with gr.TabItem("Leaderboard"):
            with gr.Row():
                data = gr.outputs.Dataframe(type="pandas")
            with gr.Row():
                data_run = gr.Button("Refresh")
                data_run.click(data_provider.fetch_data, inputs=None, outputs=data)
    # running the function on page load in addition to when the button is clicked
    block.load(data_provider.fetch_data, inputs=None, outputs=data)               

block.launch()