How to beginner · 3 min read

How to download model from Hugging Face

Q: How to download model from Hugging Face

Use the huggingface_hub Python library to download models from Hugging Face Hub. Install it via pip install huggingface_hub, then call hf_hub_download(repo_id, filename) to fetch model files programmatically.

Quick answer

Use the huggingface_hub Python library to download models from Hugging Face Hub. Install it via pip install huggingface_hub, then call hf_hub_download(repo_id, filename) to fetch model files programmatically.

PREREQUISITES

Python 3.8+
pip install huggingface_hub>=0.14.1
Hugging Face account (optional for private models)

Setup

Install the official Hugging Face Hub Python SDK to interact with the model repository. Optionally, set your Hugging Face API token as an environment variable for private model access.

bash

pip install huggingface_hub>=0.14.1

Step by step

Use the hf_hub_download function to download a specific file from a Hugging Face model repository. Provide the repository ID and filename. The function returns the local path to the downloaded file.

python

from huggingface_hub import hf_hub_download

# Download the PyTorch model file from the 'bert-base-uncased' repo
local_path = hf_hub_download(repo_id="bert-base-uncased", filename="pytorch_model.bin")

print(f"Model downloaded to: {local_path}")

output

Model downloaded to: /home/user/.cache/huggingface/hub/models--bert-base-uncased/snapshots/abcdef123456/pytorch_model.bin

Common variations

Download tokenizer files by specifying their filenames, e.g., tokenizer.json.
Use repo_type="dataset" to download dataset files.
Authenticate with use_auth_token=True for private models.
Use async versions with asyncio and huggingface_hub async API.

python

from huggingface_hub import hf_hub_download

# Download tokenizer file
local_tokenizer = hf_hub_download(repo_id="bert-base-uncased", filename="tokenizer.json")

# Download private model file with authentication
import os
local_private = hf_hub_download(
    repo_id="private-user/private-model",
    filename="pytorch_model.bin",
    use_auth_token=os.environ["HF_API_TOKEN"]
)

print(f"Tokenizer downloaded to: {local_tokenizer}")
print(f"Private model downloaded to: {local_private}")

output

Tokenizer downloaded to: /home/user/.cache/huggingface/hub/models--bert-base-uncased/snapshots/abcdef123456/tokenizer.json
Private model downloaded to: /home/user/.cache/huggingface/hub/models--private-user-private-model/snapshots/123456abcdef/pytorch_model.bin

Troubleshooting

If you get a RepositoryNotFoundError, verify the repo_id spelling and your access rights.
For authentication errors, ensure your Hugging Face token is valid and set in HF_API_TOKEN environment variable.
Check your internet connection if downloads fail or timeout.
Clear the cache folder ~/.cache/huggingface/hub if corrupted files cause issues.

✅

Key Takeaways

Use the official huggingface_hub Python SDK to download models easily.
Set your Hugging Face API token in environment variables for private model access.
Specify exact filenames when downloading specific model or tokenizer files.
Handle common errors by verifying repo IDs and authentication tokens.
Cache is stored locally under ~/.cache/huggingface/hub for reuse.

Verified 2026-04 · bert-base-uncased

Verify ↗