Spaces:
Sleeping
Sleeping
File size: 3,656 Bytes
85d3bc8 60ba391 19112df 85d3bc8 19112df 85d3bc8 60ba391 19112df 61f4e5e 85d3bc8 19112df 7b638ad 85d3bc8 19112df 85d3bc8 19112df 85d3bc8 19112df 85d3bc8 7b638ad 19112df |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 |
import gradio as gr
import pandas as pd
import requests
from datetime import datetime
from apscheduler.schedulers.background import BackgroundScheduler
from src.about import (
INTRODUCTION_TEXT,
LLM_BENCHMARKS_TEXT,
TITLE,
)
from src.display.css_html_js import custom_css
from huggingface_hub import HfApi
repo_id = "qinference/AIM100Leaderboard"
api = HfApi()
last_updated = datetime.now().strftime('Last updated at %Y-%m-%d %H:%m:%S')
def restart_leaderboard():
api.pause_space(repo_id=repo_id)
api.restart_space(repo_id=repo_id)
def get_evaluation():
global last_updated
try:
response = requests.get("http://aim100.qinference.com/api/leaderboard/list")
data_json = response.json()
df = pd.DataFrame(data_json)
for col in df.columns:
df.loc[df[col] == 0, col] = '-'
df.insert(0, 'No', df.reset_index().index + 1)
ret = df.drop(columns='nodeSeq').rename(columns={'modelName': 'Model'})
ret.columns = [x.capitalize() for x in ret.columns]
except ValueError:
ret = default_evaluation()
last_updated = "<p style='text-align: right; padding-right: 5px;'>" + datetime.now().strftime('Last updated at %Y-%m-%d %H:%m:%S') + "</p>"
return ret
def default_evaluation():
global last_updated
default_data = [{
"No": "-",
"Model": "-",
"Total": "-",
"Inference": "-",
"Grammar": "-",
"Understanding": "-",
"Coding": "-",
"Math": "-",
"Writing": "-",
"Etc": "-"
}]
df = pd.DataFrame(default_data)
last_updated = datetime.now().strftime('Last updated at %Y-%m-%d %H:%m:%S')
return df
# Searching and filtering
def update_table(
hidden_df: pd.DataFrame,
query: str,
):
filtered_df = hidden_df[hidden_df["Model"].str.contains(query, case=False)]
return filtered_df
original_df = get_evaluation()
leaderboard_df = original_df.copy()
leaderboard = gr.Blocks(css=custom_css)
with leaderboard:
gr.HTML(TITLE)
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
with gr.Tabs(elem_classes="tab-buttons") as tabs:
with gr.TabItem("🏅 LLM Benchmark", elem_id="llm-benchmark-tab-table", id=0):
with gr.Row():
search_bar = gr.Textbox(
placeholder=" 🔍 Search for your model (separate multiple queries with `;`) and press ENTER...",
show_label=False,
elem_id="search-bar",
)
leaderboard_table = gr.components.Dataframe(
value=leaderboard_df,
elem_id="leaderboard-table",
interactive=False,
visible=True,
)
# Dummy leaderboard for handling the case when the user uses backspace key
hidden_leaderboard_table_for_search = gr.components.Dataframe(
value=original_df,
visible=False,
)
search_bar.submit(
update_table,
[
hidden_leaderboard_table_for_search,
search_bar
],
leaderboard_table,
)
scheduler = BackgroundScheduler()
scheduler.add_job(restart_leaderboard, "interval", seconds=60)
scheduler.start()
with gr.Row():
gr.HTML(last_updated)
with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
leaderboard.queue(default_concurrency_limit=40).launch()
|