leaderboard / app.py
kunato's picture
add report submited model
f04ff4d
import subprocess
import gradio as gr
from apscheduler.schedulers.background import BackgroundScheduler
from huggingface_hub import snapshot_download
from src.pages.about import show_about_page
from src.pages.submit import show_submit_page
from src.pages.result_table import show_result_page
from src.about import (
CITATION_BUTTON_LABEL,
CITATION_BUTTON_TEXT,
INTRODUCTION_TEXT,
TITLE,
)
from src.display.css_html_js import custom_css
from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN
from watch_leaderboard import watch_submit_queue
def restart_space():
API.restart_space(repo_id=REPO_ID)
try:
print(EVAL_REQUESTS_PATH)
snapshot_download(
repo_id=QUEUE_REPO,
local_dir=EVAL_REQUESTS_PATH,
repo_type="dataset",
tqdm_class=None,
etag_timeout=30,
token=TOKEN,
)
except Exception:
restart_space()
try:
print(EVAL_RESULTS_PATH)
snapshot_download(
repo_id=RESULTS_REPO,
local_dir=EVAL_RESULTS_PATH,
repo_type="dataset",
tqdm_class=None,
etag_timeout=30,
token=TOKEN,
)
except Exception:
restart_space()
demo = gr.Blocks(css=custom_css)
with demo:
gr.HTML(TITLE)
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
with gr.Tabs(elem_classes="tab-buttons") as tabs:
show_result_page(root_path="MC", title="πŸ“ Exam", index=0)
show_result_page(
root_path="LLM",
title="πŸ€– LLM-as-a-Judge",
index=1,
extra=lambda: gr.Markdown(
"* Gemini result is only reported in MT-Bench because the other benchmark was mistakenly classified as unsafe"
),
)
show_result_page(root_path="NLU", title="πŸ•΅οΈ NLU", index=2)
show_result_page(root_path="NLG", title="πŸ–ŠοΈ NLG", index=3)
show_about_page(index=4)
show_submit_page(index=5)
with gr.Column():
with gr.Accordion("πŸ“™ Citation", open=False):
citation_button = gr.Textbox(
value=CITATION_BUTTON_TEXT,
label=CITATION_BUTTON_LABEL,
lines=8,
elem_id="citation-button",
show_copy_button=True,
)
scheduler = BackgroundScheduler()
scheduler.add_job(watch_submit_queue, "interval", seconds=3500)
scheduler.add_job(restart_space, "interval", seconds=3600)
scheduler.start()
demo.queue(default_concurrency_limit=40).launch()