File size: 3,707 Bytes
9ee2508 5d553fb 9ee2508 5d553fb 9ee2508 5d553fb 9ee2508 5d553fb 9ee2508 5d553fb 9ee2508 5d553fb 9ee2508 | 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 | import gradio as gr
import pandas as pd
# Import configurations and informational texts
from src.config import (
CITATIONS,
INFO_BENCHMARK_TASK,
INFO_GOTO_AI,
INFO_SCORE_CALCULATION,
INTRODUCTION_TEXT,
TITLE,
file_path,
hidden_tabs,
model_types,
)
# Import function to load leaderboard tables
from src.populate import load_tables
FIXED_COL_WIDTHS = {"Model": "260px", "Type": "90px", "Size": "70px", "Total": "90px"}
def column_widths_for(columns: list[str]) -> list[str]:
return [FIXED_COL_WIDTHS.get(col, "130px") for col in columns]
def make_filter_fn(df: pd.DataFrame):
def filter_table(search: str, sizes: list[str]) -> pd.DataFrame:
filtered = df
if sizes:
filtered = filtered[filtered["Size"].isin(sizes)]
if search:
filtered = filtered[filtered["Model"].str.contains(search, case=False, na=False)]
return filtered
return filter_table
# Create a Gradio application with block-based UI
# 'Blocks()' is used to group multiple components in a single interface
demo = gr.Blocks()
with demo:
gr.HTML(TITLE) # Display the main title of the application
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text") # Display introductory text
# Create tabs to display different leaderboard tables
with gr.Tabs(elem_classes="tab-buttons") as tabs:
tables = load_tables(file_path) # Load leaderboard data from file
for model_type in model_types:
with gr.TabItem(model_type, elem_id="llm-benchmark-tab-table", id=model_type):
for i, t in enumerate(tables): # Loop through the tables to create tabs
if (model_type, t["name"]) in hidden_tabs:
continue
with gr.TabItem(t["name"], elem_id="llm-benchmark-tab-table", id=i):
table_df = t["table"][t["table"]["Type"] == model_type]
table_df = table_df.drop(columns=t["hidden_col"], errors="ignore")
table_df = table_df.dropna(axis=1, how="all")
sizes = sorted(table_df["Size"].dropna().unique().tolist())
with gr.Row():
search_box = gr.Textbox(label="Search model", placeholder="Search by model name")
size_filter = gr.CheckboxGroup(choices=sizes, value=sizes, label="Model sizes")
table = gr.Dataframe(
value=table_df,
interactive=False,
wrap=True,
column_widths=column_widths_for(list(table_df.columns)),
)
filter_fn = make_filter_fn(table_df)
search_box.change(filter_fn, inputs=[search_box, size_filter], outputs=table)
size_filter.change(filter_fn, inputs=[search_box, size_filter], outputs=table)
# Add additional informational sections using Accordion
with gr.Row():
with gr.Accordion("📚 Benchmark Tasks", open=False):
gr.Markdown(INFO_BENCHMARK_TASK, elem_classes="markdown-text")
with gr.Row():
with gr.Accordion("🧮 Score Calculation", open=False):
gr.Markdown(INFO_SCORE_CALCULATION, elem_classes="markdown-text")
with gr.Row():
with gr.Accordion("🤝 About GoTo-AI", open=False):
gr.Markdown(INFO_GOTO_AI, elem_classes="markdown-text")
with gr.Row():
with gr.Accordion("📝 Citations", open=False):
gr.Markdown(CITATIONS, elem_classes="markdown-text")
# Run the application
demo.launch()
|