Spaces:
Running
Running
adding app
Browse files
app.py
ADDED
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import re
|
2 |
+
import streamlit as st
|
3 |
+
import requests
|
4 |
+
import pandas as pd
|
5 |
+
from io import StringIO
|
6 |
+
import plotly.graph_objs as go
|
7 |
+
|
8 |
+
|
9 |
+
def convert_markdown_table_to_dataframe(md_content):
|
10 |
+
"""
|
11 |
+
Converts a markdown table to a Pandas DataFrame, handling special characters, links,
|
12 |
+
and extracting Hugging Face URLs.
|
13 |
+
"""
|
14 |
+
cleaned_content = re.sub(r'\|\s*$', '', re.sub(r'^\|\s*', '', md_content, flags=re.MULTILINE), flags=re.MULTILINE)
|
15 |
+
df = pd.read_csv(StringIO(cleaned_content), sep="\|", engine='python')
|
16 |
+
df = df.drop(0, axis=0) # Remove first row if it's not the header
|
17 |
+
df.columns = df.columns.str.strip() # Clean column names
|
18 |
+
|
19 |
+
# Extract Model names and URLs
|
20 |
+
model_link_pattern = r'\[(.*?)\]\((.*?)\)'
|
21 |
+
df['URL'] = df['Model'].apply(lambda x: re.search(model_link_pattern, x).group(2) if re.search(model_link_pattern, x) else None)
|
22 |
+
df['Model'] = df['Model'].apply(lambda x: re.sub(model_link_pattern, r'\1', x))
|
23 |
+
return df
|
24 |
+
|
25 |
+
|
26 |
+
def create_bar_chart(df, metric):
|
27 |
+
"""
|
28 |
+
Creates and displays a bar chart for a given metric.
|
29 |
+
"""
|
30 |
+
st.write(f"### {metric} Scores")
|
31 |
+
if metric not in df.columns:
|
32 |
+
st.write(f"No data available for {metric}.")
|
33 |
+
return
|
34 |
+
|
35 |
+
sorted_df = df[['Model', metric]].dropna().sort_values(by=metric, ascending=True)
|
36 |
+
fig = go.Figure(go.Bar(
|
37 |
+
x=sorted_df[metric],
|
38 |
+
y=sorted_df['Model'],
|
39 |
+
orientation='h',
|
40 |
+
marker=dict(color=sorted_df[metric], colorscale='Inferno')
|
41 |
+
))
|
42 |
+
fig.update_layout(margin=dict(l=20, r=20, t=20, b=20))
|
43 |
+
st.plotly_chart(fig, use_container_width=True)
|
44 |
+
|
45 |
+
|
46 |
+
def main():
|
47 |
+
st.set_page_config(page_title="LLM Leaderboard", layout="wide")
|
48 |
+
st.title("🏆 LLM Leaderboard")
|
49 |
+
|
50 |
+
# URL to your markdown file
|
51 |
+
md_url = st.text_input("Enter the URL to the markdown file", "https://raw.githubusercontent.com/yourrepo/README.md")
|
52 |
+
|
53 |
+
if not md_url:
|
54 |
+
st.error("Please provide a valid URL to a markdown file containing the leaderboard table.")
|
55 |
+
return
|
56 |
+
|
57 |
+
try:
|
58 |
+
response = requests.get(md_url)
|
59 |
+
response.raise_for_status()
|
60 |
+
md_content = response.text
|
61 |
+
|
62 |
+
df = convert_markdown_table_to_dataframe(md_content)
|
63 |
+
|
64 |
+
# Automatically detect metrics (all columns except 'Model' and 'URL')
|
65 |
+
metric_columns = [col for col in df.columns if col not in ['Model', 'URL']]
|
66 |
+
|
67 |
+
# Convert metric columns to numeric, handling errors gracefully
|
68 |
+
for col in metric_columns:
|
69 |
+
df[col] = pd.to_numeric(df[col], errors='coerce')
|
70 |
+
|
71 |
+
# Sortable leaderboard table
|
72 |
+
st.dataframe(
|
73 |
+
df[['Model'] + metric_columns + ['URL']],
|
74 |
+
use_container_width=True,
|
75 |
+
hide_index=True,
|
76 |
+
)
|
77 |
+
|
78 |
+
# Bar charts for each metric
|
79 |
+
for metric in metric_columns:
|
80 |
+
create_bar_chart(df, metric)
|
81 |
+
|
82 |
+
except Exception as e:
|
83 |
+
st.error(f"An error occurred while processing the markdown table: {e}")
|
84 |
+
|
85 |
+
|
86 |
+
if __name__ == "__main__":
|
87 |
+
main()
|
88 |
+
|