Spaces:
Running
Running
| import pandas as pd | |
| from pathlib import Path | |
| from datasets import load_dataset | |
| import numpy as np | |
| import os | |
| import re | |
| # From Open LLM Leaderboard | |
| def model_hyperlink(link, model_name): | |
| if model_name == "random": | |
| return "random" | |
| return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{model_name}</a>' | |
| def undo_hyperlink(html_string): | |
| # Regex pattern to match content inside > and < | |
| pattern = r'>[^<]+<' | |
| match = re.search(pattern, html_string) | |
| if match: | |
| # Extract the matched text and remove leading '>' and trailing '<' | |
| return match.group(0)[1:-1] | |
| else: | |
| return "No text found" | |
| # Define a function to fetch and process data | |
| def load_all_data(data_repo, subdir:str, subsubsets=False): # use HF api to pull the git repo | |
| dir = Path(data_repo) | |
| data_dir = dir / subdir | |
| orgs = [d for d in os.listdir(data_dir) if os.path.isdir(os.path.join(data_dir, d))] | |
| # get all files within the sub folders orgs | |
| models_results = [] | |
| for org in orgs: | |
| org_dir = data_dir / org | |
| files = [f for f in os.listdir(org_dir) if os.path.isfile(os.path.join(org_dir, f))] | |
| for file in files: | |
| if file.endswith(".json"): | |
| models_results.append(org + "/" + file) | |
| # create empty dataframe to add all data to | |
| df = pd.DataFrame() | |
| # load all json data in the list models_results one by one to avoid not having the same entries | |
| for model in models_results: | |
| model_data = load_dataset("json", data_files=data_repo + subdir+ "/" + model, split="train") | |
| df2 = pd.DataFrame(model_data) | |
| # add to df | |
| df = pd.concat([df2, df]) | |
| # remove chat_template comlumn | |
| df = df.drop(columns=["chat_template"]) | |
| # sort columns alphabetically | |
| df = df.reindex(sorted(df.columns), axis=1) | |
| # move column "model" to the front | |
| cols = list(df.columns) | |
| cols.insert(0, cols.pop(cols.index('model'))) | |
| df = df.loc[:, cols] | |
| # select all columns except "model" | |
| cols = df.columns.tolist() | |
| cols.remove("model") | |
| # round | |
| df[cols] = df[cols].round(2) | |
| avg = np.nanmean(df[cols].values,axis=1).round(2) | |
| # add average column | |
| df["average"] = avg | |
| # apply model_hyperlink function to column "model" | |
| df["model"] = df["model"].apply(lambda x: model_hyperlink(f"https://huggingface.co/{x}", x)) | |
| # move average column to the second | |
| cols = list(df.columns) | |
| cols.insert(1, cols.pop(cols.index('average'))) | |
| df = df.loc[:, cols] | |
| # remove column xstest (outdated data) | |
| # if xstest is a column | |
| if "xstest" in df.columns: | |
| df = df.drop(columns=["xstest"]) | |
| # remove column anthropic and summarize_prompted (outdated data) | |
| if "anthropic" in df.columns: | |
| df = df.drop(columns=["anthropic"]) | |
| if "summarize_prompted" in df.columns: | |
| df = df.drop(columns=["summarize_prompted"]) | |
| return df | |