app.py

import gradio as gr
import json
import pandas as pd
from urllib.request import urlopen
from urllib.error import URLError
import re
from datetime import datetime

CITATION_BUTTON_TEXT = r"""@misc{2023opencompass,
    title={OpenCompass: A Universal Evaluation Platform for Foundation Models},
    author={OpenCompass Contributors},
    howpublished = {\url{https://github.com/open-compass/opencompass}},
    year={2023}
}"""
CITATION_BUTTON_LABEL = "Copy the following snippet to cite these results"


head_style = """
<style>
@media (min-width: 1536px)
{
    .gradio-container {
        min-width: var(--size-full) !important;
    }
}
</style>
"""


DATA_URL_BASE = "http://opencompass.oss-cn-shanghai.aliyuncs.com/dev-assets/hf-research/"

def findfile():
    model_meta_info = 'model-meta-info'
    results_sum = 'hf-academic'

    url = f"{DATA_URL_BASE}{model_meta_info}.json"
    response = urlopen(url)
    model_info = json.loads(response.read().decode('utf-8'))

    url = f"{DATA_URL_BASE}{results_sum}.json"
    response = urlopen(url)
    results = json.loads(response.read().decode('utf-8'))

    return model_info, results


MAIN_LEADERBOARD_DESCRIPTION = """## Compass Academic Leaderboard
--WIP--

"""
Initial_title = 'Compass Academic Leaderboard'



def make_results_tab(model_info, results):
    models_list, datasets_list = [], []
    for i in model_info:
        models_list.append([i['abbr'], i['display_name']])
    for i in results.keys():
        datasets_list.append(i)
    
    result_list = []
    index = 0
    for model in models_list:
        this_result = {}
        this_result['Index'] = index
        this_result['Model Name'] = model[1]
        index += 1        
        for dataset in datasets_list:
            this_result[dataset] = results[dataset][model[0]]
        result_list.append(this_result)

    df = pd.DataFrame(result_list)
    return df 


def calculate_column_widths(df):
    column_widths = []
    for column in df.columns:
        header_length = len(str(column))
        max_content_length = df[column].astype(str).map(len).max()
        width = max(header_length * 10, max_content_length * 8) + 20
        width = max(160, min(400, width))
        column_widths.append(width)
    return column_widths


def show_results_tab(df, model_info, results):
    
    def filter_df(model_name):
        df = make_results_tab(model_info, results)
        default_val = 'Input the Model Name (fuzzy)'
        if model_name != default_val:
            method_names = [x.split('</a>')[0].split('>')[-1].lower() for x in df['Model Name']]
            flag = [model_name.lower() in name for name in method_names]
            df['TEMP'] = flag
            df = df[df['TEMP'] == True] 
            df.pop('TEMP')
    return df

    with gr.Row():
        model_name = gr.Textbox(
            value='Input the Model Name (fuzzy)', 
            label='Search Model Name',
            interactive=True
        )


    
    table = gr.DataFrame(
            value=df,
            interactive=False,
            wrap=False,
            column_widths=calculate_column_widths(df),
    )

    # search_box.submit(
    #     fn=filter_df,
    #     inputs=search_box,
    #     outputs=table
    # )



def create_interface():
    model_info, results = findfile()

    with gr.Blocks() as demo:
        # title_comp = gr.Markdown(Initial_title)
        gr.Markdown(MAIN_LEADERBOARD_DESCRIPTION)
        with gr.Tabs(elem_classes='tab-buttons') as tabs:
            with gr.TabItem('Results', elem_id='main', id=0):
                df = make_results_tab(model_info, results)
                show_results_tab(df, model_info, results)

            with gr.TabItem('Predictions', elem_id='notmain', id=1):
                # dataset_tab(results, structs[i], dataset)
                pass

    return demo

# model_info, results = findfile()
# breakpoint()

if __name__ == '__main__':
    demo = create_interface()
    demo.queue()
    demo.launch(server_name='0.0.0.0')