what is hugging face

Hugging Face

https://www.youtube.com/watch?v=GLO5FZzfrS0

Lets try to understand what is what is hugging face as quick as possible.

• Hugging Face is an open-source platform and community often described as the GitHub of machine learning,

• This help to share, build, and deploy AI models, datasets, and applications.

• Also known as Transformers library

• Its also hosting datasets.

• Hugging Face Hub: A central repository for sharing and discovering models, datasets, and spaces.

• Spaces: A hosting service for creating and showcasing interactive ML demos and applications.

• Collaborative Ecosystem: A vibrant community of researchers and developers contributing to open-source AI innovation.

• Integration: Easy integration with popular frameworks like PyTorch and TensorFlow via its Python libraries.

Installing hugging face

# Recommended standalone installer
curl -LsSf https://hf.co/cli/install.sh | bash

# Or via pip
pip install -U "huggingface_hub"   

Few quick command

Authentication: hf auth login (opens browser) or hf auth login --token $HF_TOKEN.

Upload: hf upload <repo_id> <local_file_or_folder> (replaces git push).

Download: hf download <repo_id> (replaces git clone for single files/folders).

Repo Management: hf repo create <repo_id> --type model (creates a new repository).

Sync: hf sync (syncs local directories with remote buckets/repos).

2. Native Git Integration
Because every Hugging Face repository is a Git repository (backed by Git Large File Storage or Xet for large models), you can use standard git commands exactly as you would with GitHub. 

Clone:
git clone https://huggingface.co/<username>/<repo_name>
# Or via SSH
git clone [email protected]:<username>/<repo_name>

Push/Pull:
git add .
git commit -m "Update model weights"
git push

Setup: To enable password-less pushes, run hf auth login --add-to-git-credential. This saves your token to your system's git credential helper, so standard git push commands authenticate automatically. 

• Inference Endpoints: A fully managed service to deploy models on dedicated infrastructure (CPU or GPU) across AWS, Azure, or GCP. It handles scaling, security, and monitoring, allowing developers to turn a model into a production API in minutes rather than days.

• Inference Providers: A serverless gateway that allows users to run models via third-party providers without managing infrastructure, often charging only per token or request.