Introduction to Huggingface Hub

Huggingface Hub is a platform that provides a comprehensive set of tools and APIs to build, train, and deploy machine learning models. It enables seamless collaboration among developers, data scientists, and researchers by offering a centralized repository for model sharing and discovery.

Useful API Examples

1. Creating a Model Repository

  from huggingface_hub import HfApi
  
  api = HfApi()
  repo_url = api.create_repo(name="my-awesome-model", private=True)
  print(f"Model repository created at: {repo_url}")

2. Uploading a Model

  from huggingface_hub import upload_file
  
  upload_file(path_or_fileobj="pytorch_model.bin", path_in_repo="my-awesome-model/pytorch_model.bin", repo_id="username/my-awesome-model")
  print("Model uploaded successfully!")

3. Downloading a Model

  from huggingface_hub import hf_hub_download
  
  model_path = hf_hub_download(repo_id="username/my-awesome-model", filename="pytorch_model.bin")
  print(f"Model downloaded to: {model_path}")

4. Listing All Repositories of a User

  from huggingface_hub import HfApi
  
  api = HfApi()
  user_repos = api.list_repos(username="username")
  for repo in user_repos:
      print(repo)

5. Deleting a Model Repository

  from huggingface_hub import HfApi
  
  api = HfApi()
  api.delete_repo(name="my-awesome-model", username="username")
  print("Model repository deleted successfully.")

Building an App with Huggingface Hub APIs

Below is a simple example of a web application that utilizes Huggingface Hub APIs to manage model repositories. The app enables users to upload, download, and delete models from the Hub.

App Example

  from flask import Flask, request, jsonify
  from huggingface_hub import HfApi, upload_file, hf_hub_download

  app = Flask(__name__)
  api = HfApi()

  @app.route('/create_repo', methods=['POST'])
  def create_repo():
      data = request.json
      repo_url = api.create_repo(name=data['name'], private=data['private'])
      return jsonify({"repo_url": repo_url})

  @app.route('/upload_model', methods=['POST'])
  def upload_model():
      data = request.json
      upload_file(path_or_fileobj=data['file_path'], path_in_repo=f"{data['repo_name']}/{data['file_name']}", repo_id=data['repo_id'])
      return jsonify({"status": "Model uploaded successfully!"})

  @app.route('/download_model', methods=['GET'])
  def download_model():
      repo_id = request.args.get('repo_id')
      filename = request.args.get('filename')
      model_path = hf_hub_download(repo_id=repo_id, filename=filename)
      return jsonify({"model_path": model_path})

  @app.route('/delete_repo', methods=['DELETE'])
  def delete_repo():
      data = request.json
      api.delete_repo(name=data['name'], username=data['username'])
      return jsonify({"status": "Model repository deleted successfully."})

  if __name__ == '__main__':
      app.run(debug=True)

This example showcases the integration of Huggingface Hub’s functionalities into a Flask web application, highlighting the ease of use and flexibility of the APIs.

Hash: 61f652e7088dbd0a6d0f59f18265ba2aef2f1d5a8c7715277dd31ff45e01fcb4