|
|
|
|
|
|
|
|
|
import shutil |
|
|
|
import gradio as gr |
|
from apscheduler.schedulers.background import BackgroundScheduler |
|
from gradio_leaderboard import ColumnFilter, Leaderboard, SelectColumns |
|
from huggingface_hub import snapshot_download |
|
|
|
from src.about import ( |
|
CITATION_BUTTON_LABEL, |
|
CITATION_BUTTON_TEXT, |
|
EVALUATION_REQUESTS_TEXT, |
|
EVALUATION_SCRIPT, |
|
INTRODUCTION_TEXT, |
|
LLM_BENCHMARKS_TEXT, |
|
TITLE, |
|
) |
|
from src.display.css_html_js import custom_css |
|
from src.display.utils import ( |
|
BENCHMARK_COLS, |
|
COLS, |
|
EVAL_COLS, |
|
EVAL_TYPES, |
|
AutoEvalColumn, |
|
ModelType, |
|
Precision, |
|
WeightType, |
|
fields, |
|
) |
|
from src.envs import ( |
|
API, |
|
CACHE_PATH, |
|
EVAL_REQUESTS_PATH, |
|
EVAL_RESULTS_PATH, |
|
REPO_ID, |
|
REQUESTS_REPO, |
|
RESULTS_REPO, |
|
TOKEN, |
|
) |
|
from src.populate import get_evaluation_requests_df, get_leaderboard_df |
|
from src.submission.submit import add_new_eval |
|
|
|
|
|
def restart_space(): |
|
API.restart_space(repo_id=REPO_ID) |
|
|
|
|
|
shutil.rmtree(CACHE_PATH, ignore_errors=True) |
|
try: |
|
snapshot_download( |
|
repo_id=REQUESTS_REPO, |
|
local_dir=EVAL_REQUESTS_PATH, |
|
repo_type="dataset", |
|
tqdm_class=None, |
|
etag_timeout=30, |
|
token=TOKEN, |
|
) |
|
except Exception: |
|
restart_space() |
|
|
|
try: |
|
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() |
|
|
|
LEADERBOARD_DF = get_leaderboard_df( |
|
EVAL_RESULTS_PATH, |
|
EVAL_REQUESTS_PATH, |
|
COLS, |
|
BENCHMARK_COLS, |
|
) |
|
|
|
( |
|
finished_eval_requests_df, |
|
running_eval_requests_df, |
|
pending_eval_requests_df, |
|
) = get_evaluation_requests_df(EVAL_REQUESTS_PATH, EVAL_COLS) |
|
|
|
def init_leaderboard(dataframe): |
|
if dataframe is None or dataframe.empty: |
|
raise ValueError("Leaderboard DataFrame is empty or None.") |
|
return Leaderboard( |
|
value=dataframe, |
|
datatype=[c.type for c in fields(AutoEvalColumn)], |
|
select_columns=SelectColumns( |
|
default_selection=[c.name for c in fields(AutoEvalColumn) if c.displayed_by_default], |
|
cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden], |
|
label="Columns", |
|
), |
|
search_columns=[AutoEvalColumn.model.name, AutoEvalColumn.license.name], |
|
hide_columns=[c.name for c in fields(AutoEvalColumn) if c.hidden], |
|
filter_columns=[ |
|
ColumnFilter(AutoEvalColumn.precision.name, type="checkboxgroup", label="Floating-point format"), |
|
ColumnFilter( |
|
AutoEvalColumn.params.name, |
|
type="slider", |
|
min=1, |
|
max=500, |
|
label="Number of parameters (billions)", |
|
), |
|
], |
|
interactive=False, |
|
) |
|
|
|
|
|
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: |
|
with gr.TabItem("π Ranking", elem_id="llm-benchmark-tab-table", id=0): |
|
leaderboard = init_leaderboard(LEADERBOARD_DF) |
|
|
|
with gr.TabItem("π§ About", elem_id="llm-benchmark-tab-table", id=2): |
|
gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text") |
|
with gr.Accordion( |
|
"Evaluation script", |
|
open=False, |
|
): |
|
gr.Markdown( |
|
EVALUATION_SCRIPT, |
|
elem_classes="markdown-text", |
|
) |
|
|
|
with gr.TabItem("π§ͺ Submissions", elem_id="llm-benchmark-tab-table", id=3): |
|
with gr.Column(): |
|
with gr.Row(): |
|
gr.Markdown(EVALUATION_REQUESTS_TEXT, elem_classes="markdown-text") |
|
|
|
with gr.Column(): |
|
with gr.Accordion( |
|
f"β
Finished ({len(finished_eval_requests_df)})", |
|
open=False, |
|
): |
|
with gr.Row(): |
|
finished_eval_table = gr.components.Dataframe( |
|
value=finished_eval_requests_df, |
|
headers=EVAL_COLS, |
|
datatype=EVAL_TYPES, |
|
row_count=5, |
|
) |
|
|
|
with gr.Accordion( |
|
f"β³ Pending ({len(pending_eval_requests_df)})", |
|
open=False, |
|
): |
|
with gr.Row(): |
|
pending_eval_table = gr.components.Dataframe( |
|
value=pending_eval_requests_df, |
|
headers=EVAL_COLS, |
|
datatype=EVAL_TYPES, |
|
row_count=5, |
|
) |
|
with gr.Row(): |
|
gr.Markdown("# βοΈ Submit a model", elem_classes="markdown-text") |
|
|
|
with gr.Row(): |
|
with gr.Column(): |
|
model_name_textbox = gr.Textbox(label="Model name") |
|
revision_name_textbox = gr.Textbox(label="Revision commit", placeholder="main") |
|
model_type = gr.Dropdown( |
|
choices=[t.to_str(" ") for t in ModelType if t not in [ModelType.Unknown, ModelType.BrainDAO]], |
|
label="Model type", |
|
multiselect=False, |
|
value=None, |
|
interactive=True, |
|
) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
submit_button = gr.Button("Submit") |
|
submission_result = gr.Markdown() |
|
submit_button.click( |
|
add_new_eval, |
|
[ |
|
model_name_textbox, |
|
|
|
revision_name_textbox, |
|
|
|
|
|
model_type, |
|
], |
|
submission_result, |
|
) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
scheduler = BackgroundScheduler() |
|
scheduler.add_job(restart_space, "interval", seconds=3600) |
|
scheduler.start() |
|
demo.queue(default_concurrency_limit=40).launch( |
|
server_name="0.0.0.0", |
|
allowed_paths=["images/solbench.svg"], |
|
) |
|
|