Spaces:
Runtime error
Runtime error
| import gradio as gr | |
| import pandas as pd | |
| import json | |
| from datasets import load_dataset | |
| import requests | |
| from huggingface_hub import list_datasets, list_models, list_spaces | |
| from collections import Counter | |
| import numpy as np | |
| def compute_ranking(df, column, method="sum", keep="last"): | |
| df_rank = df.groupby("author").aggregate({column: method})[[column]] | |
| df_rank = df_rank.sort_values(by=column) | |
| df_rank.reset_index(drop=True, inplace=True) | |
| df_rank["top_perc"] = df_rank.apply(lambda x: f"{100 * (1-(x.name/len(df_rank))):.2f}", axis=1) | |
| df_rank = df_rank.drop_duplicates(subset=column, keep=keep) | |
| df_rank = df_rank.rename({column: "value"}, axis='columns') | |
| return df_rank | |
| class NpEncoder(json.JSONEncoder): | |
| def default(self, obj): | |
| if isinstance(obj, np.integer): | |
| return int(obj) | |
| if isinstance(obj, np.floating): | |
| return float(obj) | |
| if isinstance(obj, np.ndarray): | |
| return obj.tolist() | |
| return super(NpEncoder, self).default(obj) | |
| ds = load_dataset("open-source-metrics/model-repos-stats", split="train") | |
| df = ds.to_pandas() | |
| df_ranks = {} | |
| df_ranks["likes"] = compute_ranking(df, "likes") | |
| df_ranks["downloads"] = compute_ranking(df, "downloads_30d") | |
| df_ranks["repos"] = compute_ranking(df, "repo_id", method="count") | |
| with open("./html_template.html", "r") as f: | |
| template = f.read() | |
| def create_user_summary(user_name): | |
| summary = {} | |
| df_user = df.loc[df["author"]==user_name] | |
| if len(df_user) == 0: | |
| return """<br><p style="text-align: center;color: rgb(255, 210, 31);font-family: 'Consolas', monospace; font-size: 24px;">Unfortunately there is not enough data for your report.</p><br>""" | |
| r = requests.get(f'https://huggingface.co/api/users/{user_name}/likes') | |
| user_datasets = [dataset for dataset in list_datasets(author=user_name)] | |
| user_spaces = [space for space in list_spaces(author=user_name)] | |
| summary["likes_user_total"] = df_user["likes"].sum() | |
| summary["likes_user_given"] = len(r.json()) | |
| summary["likes_user_top"] = df_ranks["likes"][df_ranks["likes"]["value"]>=summary["likes_user_total"]].iloc[0]["top_perc"] | |
| summary["likes_repo_most"] = df_user.sort_values(by="likes", ascending=False).iloc[0]["repo_id"] | |
| summary["likes_repo_most_n"] = df_user.sort_values(by="likes", ascending=False).iloc[0]["likes"] | |
| summary["downloads_user_total"] = df_user["downloads_30d"].sum() | |
| summary["downloads_user_top"] = df_ranks["downloads"][df_ranks["downloads"]["value"]>=summary["downloads_user_total"]].iloc[0]["top_perc"] | |
| summary["downlods_repo_most"] = df_user.sort_values(by="downloads_30d", ascending=False).iloc[0]["repo_id"] | |
| summary["downlods_repo_most_n"] = df_user.sort_values(by="downloads_30d", ascending=False).iloc[0]["downloads_30d"] | |
| summary["repos_model_total"] = len(df_user) | |
| summary["repos_model_top"] = df_ranks["repos"][df_ranks["repos"]["value"]>=summary["repos_model_total"]].iloc[0]["top_perc"] | |
| summary["repos_model_fav_type"] = Counter(df_user["model_type"].dropna()).most_common(1)[0][0] | |
| summary["repos_datasets_total"] = len(user_datasets) | |
| summary["repos_spaces_total"] = len(user_spaces) | |
| summary["repos_spaces_fav_sdk"] = Counter([getattr(info, "sdk", "n/a") for info in user_spaces]).most_common(1)[0][0] | |
| return dict_to_html(summary) | |
| def dict_to_html(summary): | |
| report = template | |
| for key in summary: | |
| report = report.replace("{{" + key + "}}", str(summary[key])) | |
| return report | |
| demo = gr.Blocks( | |
| css=".gradio-container {background-color: #000000}" | |
| ) | |
| with demo: | |
| with gr.Row(): | |
| gr.HTML(value="""<p style="text-align: center; color: rgb(255, 210, 31); font-family: 'Consolas', monospace; font-size: 24px;"> <b>Enter your HF user name:</b></p>""") | |
| with gr.Row(): | |
| username = gr.Textbox(lines=1, max_lines=1, label="User name") | |
| with gr.Row(): | |
| run = gr.Button() | |
| with gr.Row(): | |
| output = gr.HTML(label="Generated code") | |
| event = run.click(create_user_summary, [username], output) | |
| demo.launch() |