File size: 661 Bytes
8c49cb6
 
117d89c
7fdb5f5
8c49cb6
 
7fdb5f5
 
 
c1b8a96
3dfaf22
b1a1395
8c49cb6
b1a1395
8c49cb6
 
 
b1a1395
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
import pandas as pd

from src.display.utils import AutoEvalColumn
from src.leaderboard.read_evals import get_raw_eval_results, EvalResult


def get_leaderboard_df(
    results_path: str, requests_path: str, cols: list, benchmark_cols: list
) -> tuple[list[EvalResult], pd.DataFrame]:
    """Creates a dataframe from all the individual experiment results"""
    raw_data = get_raw_eval_results(results_path, requests_path)
    all_data_json = [v.to_dict() for v in raw_data]

    df = pd.DataFrame.from_records(all_data_json)
    df = df.sort_values(by=[AutoEvalColumn.average.name], ascending=False)
    df = df[cols].round(decimals=2)

    return raw_data, df