Using Phi Family in Hugging Face

May 1, 2025 · View on GitHub

Hugging Face is a very popular AI community with rich data and open source model resources. Different manufacturers will release open source LLM and SLM through Hugging Face, such as Microsoft, Meta, Mistral, Apple, Google, etc.

Microsoft Phi Family has been released on Hugging Face. Developers can download the corresponding Phi Family's model based on scenarios and businesses。In addition to deploying Phi Pytorch models on Hugging Face, we also released quantized models, using GGUF and ONNX formats to give end users a choice.

Download Models in Hugging face

You can download Phi family model with this link

Microsoft Modelson Hugging Face

You can download the model in different ways, such as installing the Hugging face CLI SDK or use git clone.

Using Hugginng face CLI to Download Phi Family model

  • Install Hugging face CLI

pip install -U "huggingface_hub[cli]"
  • Using huggingface-cli to login

Login to Hugging face with User Access Token from your Settings page


huggingface-cli login --token $HF_TOKEN --add-to-git-credential
  • Download

You can download model and save it to cache


huggingface-cli download microsoft/phi-4

You can set location in your special location


huggingface-cli download microsoft/phi-4 --local-dir $YOUR_PATH

Using git clone to Download Phi Family model

You can use git clone to download model too


git lfs install

git clone https://huggingface.co/microsoft/phi-4

Samples - Inference Microsoft Phi-4

  • Installing transformers library

pip install transformers -U
  • Running this code in VSCode

import transformers

pipeline = transformers.pipeline(
    "text-generation",
    model="microsoft/phi-4",
    model_kwargs={"torch_dtype": "auto"},
    device_map="auto",
)

messages = [
    {"role": "user", "content": "I have \$20,000 in my savings account, where I receive a 4% profit per year and payments twice a year. Can you please tell me how long it will take for me to become a millionaire? Also, can you please explain the math step by step as if you were explaining it to an uneducated person?"},
]

outputs = pipeline(messages, max_new_tokens=2048)
print(outputs[0]["generated_text"][-1])