app.py

# some code blocks are taken from https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard/tree/main
import json
import os
from datetime import datetime, timezone

import gradio as gr
import pandas as pd

from src.css_html import custom_css
from src.text_content import ABOUT_TEXT, SUBMISSION_TEXT_3
from src.utils import (
    AutoEvalColumn,
    fields,
    is_model_on_hub,
    make_clickable_names,
)

df = pd.read_csv("data/code_eval_board.csv")

COLS = [c.name for c in fields(AutoEvalColumn) if not c.hidden]
TYPES = [c.type for c in fields(AutoEvalColumn) if not c.hidden]
COLS_LITE = [
    c.name for c in fields(AutoEvalColumn) if c.displayed_by_default and not c.hidden
]
TYPES_LITE = [
    c.type for c in fields(AutoEvalColumn) if c.displayed_by_default and not c.hidden
]


def select_columns(df, columns):
    always_here_cols = [
        AutoEvalColumn.model.name,
    ]
    # We use COLS to maintain sorting
    filtered_df = df[
        always_here_cols + [c for c in COLS if c in df.columns and c in columns]
    ]
    return filtered_df


def filter_items(df, leaderboard_table, query):
    if query == "all":
        return df[leaderboard_table.columns]
    else:
        query = query[0]
    filtered_df = df[df["T"].str.contains(query, na=False)]
    return filtered_df[leaderboard_table.columns]


def search_table(df, leaderboard_table, query):
    filtered_df = df[(df["Model"].str.contains(query, case=False))]
    return filtered_df[leaderboard_table.columns]


df = make_clickable_names(df)

demo = gr.Blocks(css=custom_css)
with demo:
    with gr.Row():
        gr.Markdown(
            """<div style="text-align: center;"><h1> ESPnet-EZ Leaderboard for LibriSpeech-100h ASR1</span></h1></div>\
            <br>\
            <p>Users can use <code>reproduce</code> function to reproduce the numbers in ESPnet-EZ!</p>
""",
            elem_classes="markdown-text",
        )

    with gr.Tabs(elem_classes="tab-buttons") as tabs:
        with gr.TabItem("🔍 Evaluation table", id=0):
            with gr.Accordion("➡️ See All Columns", open=False):
                shown_columns = gr.CheckboxGroup(
                    choices=[
                        c
                        for c in COLS
                        if c
                        not in [
                            # AutoEvalColumn.dummy.name,
                            AutoEvalColumn.model.name,
                        ]
                    ],
                    value=[
                        c
                        for c in COLS_LITE
                        if c
                        not in [
                            # AutoEvalColumn.dummy.name,
                            AutoEvalColumn.model.name,
                        ]
                    ],
                    label="",
                    elem_id="column-select",
                    interactive=True,
                )
            # with gr.Column(min_width=780):
            with gr.Row():
                search_bar = gr.Textbox(
                    placeholder="🔍 Search for your model and press ENTER...",
                    show_label=False,
                    elem_id="search-bar",
                )
            leaderboard_df = gr.components.Dataframe(
                value=df[
                    [
                        AutoEvalColumn.model.name,
                    ]
                    + shown_columns.value
                ],
                headers=[
                    AutoEvalColumn.model.name,
                ]
                + shown_columns.value,
                datatype=TYPES,
                elem_id="leaderboard-table",
                interactive=False,
            )

            hidden_leaderboard_df = gr.components.Dataframe(
                value=df,
                headers=COLS,
                datatype=["str" for _ in range(len(COLS))],
                visible=False,
            )
            search_bar.submit(
                search_table,
                [hidden_leaderboard_df, leaderboard_df, search_bar],
                leaderboard_df,
            )
            shown_columns.change(
                select_columns,
                [hidden_leaderboard_df, shown_columns],
                leaderboard_df,
            )
            gr.Markdown(
                """
            **Notes:**
            - Win Rate represents how often a model outperforms other models in each language, averaged across all languages.
            - The scores of instruction-tuned models might be significantly higher on humaneval-python than other languages. We use the instruction format of HumanEval. For other languages, we use base MultiPL-E prompts.
            - For more details check the 📝 About section.
            - Models with a 🔴 symbol represent external evaluation submission, this means that we didn't verify the results, you can find the author's submission under `Submission PR` field from `See All Columns` tab.
            """,
                elem_classes="markdown-text",
            )

        with gr.TabItem("📝 About", id=2):
            gr.Markdown(ABOUT_TEXT, elem_classes="markdown-text")

        with gr.TabItem("Submit results 🚀", id=3):
            gr.Markdown(SUBMISSION_TEXT_3)


demo.launch()