src/build.py

import pandas as pd


df = pd.read_csv("data/raw_scores.csv")
COLS = df.columns.to_list()
# add column models_query with same values a smodels at the end of columns
df.insert(len(COLS), "models_query", df["Models"])

# average score
mean_columns = df.iloc[:,[5] + list(range(7, len(df.columns) - 3))]
df.insert(2, "Average score", mean_columns.mean(axis=1).round(2))

# add win rate columns for each language
old_size = len(df.columns)

for col in df.columns[6:-2]:
    df[col + " rank"] = df[col].rank(ascending=False)
    df[col + " rank"] = len(df) - (df[col + " rank"] - 1)
df["Win Rate"] = df.iloc[:, old_size:].mean(axis=1).round(2)
df = df.drop(df.columns[old_size:-1], axis=1)
df = df[["Models", "Size (B)", "Win Rate"] + df.columns[2:-1].tolist()]

# sort with regard to column win rate
df = df.sort_values(by=["Win Rate"], ascending=False)
# add column with model links as https://huggingface.co/WizardLM/WizardCoder-15B-V1.0, https://huggingface.co/bigcode/starcoder, https://huggingface.co/bigcode/starcoderbase, https://huggingface.co/bigcode/starcoderbase-7b,
# https://huggingface.co/bigcode/starcoderbase-3b, https://huggingface.co/bigcode/starcoderbase-1b, https://huggingface.co/bigcode/santacoder, https://huggingface.co/replit/replit-code-v1-3b, https://huggingface.co/THUDM/codegeex2-6b

links = {
    "WizardCoder-15B-V1.0": "https://huggingface.co/WizardLM/WizardCoder-15B-V1.0",
    "OctoCoder-15B": "https://huggingface.co/bigcode/octocoder",
    "OctoGeeX-7B": "https://huggingface.co/bigcode/octogeex",
    "StableCode-3B": "https://huggingface.co/stabilityai/stablecode-completion-alpha-3b",
    "StarCoder-15B": "https://huggingface.co/bigcode/starcoder",
    "StarCoderBase-15B": "https://huggingface.co/bigcode/starcoderbase",
    "StarCoderBase-7B": "https://huggingface.co/bigcode/starcoderbase-7b",
    "StarCoderBase-3B": "https://huggingface.co/bigcode/starcoderbase-3b",
    "StarCoderBase-1.1B": "https://huggingface.co/bigcode/starcoderbase-1b",
    "SantaCoder-1.1B": "https://huggingface.co/bigcode/santacoder",
    "Replit-2.7B": "https://huggingface.co/replit/replit-code-v1-3b",
    "CodeGeex2-6B": "https://huggingface.co/THUDM/codegeex2-6b",
    "CodeGen25-7B-multi": "https://huggingface.co/Salesforce/codegen25-7b-multi",
    "CodeGen25-7B-mono": "https://huggingface.co/Salesforce/codegen25-7b-mono",
    "CodeGen-16B-Multi": "https://huggingface.co/Salesforce/codegen-16B-multi",
}
df["Links"] = df["Models"].map(links)

df.insert(0, "T", "🟢")
df.loc[(df["Models"].str.contains("WizardCoder") | df["Models"].str.contains("Octo")), "T"] = "🔶"
# print first 5 rows and 10 cols
print(df.iloc[:5, :10])
df.to_csv("data/code_eval_board.csv", index=False)