Spaces:
Sleeping
Sleeping
File size: 2,048 Bytes
a3c4484 |
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 |
import gradio as gr
import pandas as pd
from glob import glob
csv_results = glob("results/*.csv")
# load the csv files into a dict with keys being name of the file and values being the data
data = {file: pd.read_csv(file) for file in csv_results}
def calculate_accuracy(df):
return df["parsed_judge_response"].mean() * 100
def accuracy_breakdown(df):
# 4 level accuracy
return (df.groupby("difficulty_level")["parsed_judge_response"].mean() * 100).values
# Define the column names with icons
headers_with_icons = [
"🤖 Model Name",
"⭐ Overall",
"📈 Level 1",
"🔍 Level 2",
"📘 Level 3",
"🔬 Level 4",
]
accuracy = {file: calculate_accuracy(data[file]) for file in data}
# Create a list to hold the data
data_for_df = []
# Define the column names with icons
# Iterate over each file and its corresponding DataFrame in the data dictionary
for file, df in data.items():
# Get the overall accuracy and round it
overall_accuracy = round(calculate_accuracy(df), 2)
# Get the breakdown accuracy and round each value
breakdown_accuracy = [round(acc, 2) for acc in accuracy_breakdown(df)]
# Prepare the model name from the file name
model_name = file.split("/")[-1].replace(".csv", "") # Corrected the file extension
# Append the data to the list
data_for_df.append([model_name, overall_accuracy] + breakdown_accuracy)
# Define the column names, adjust based on the number of difficulty levels you have
column_names = [
"Model Name",
"Overall Accuracy",
"Level 1 Accuracy",
"Level 2 Accuracy",
"Level 3 Accuracy",
"Level 4 Accuracy",
]
# Create the DataFrame
accuracy_df = pd.DataFrame(data_for_df, columns=column_names)
accuracy_df.columns = headers_with_icons
accuracy_df.sort_values(by="⭐ Overall", ascending=False, inplace=True)
with gr.Blocks() as demo:
gr.Markdown("# FSMBench Leaderboard")
# add link to home page and dataset
leader_board = gr.Dataframe(accuracy_df, headers=headers_with_icons)
demo.launch()
|