How to beginner · 3 min read

How to push fine-tuned model to Hugging Face hub

Q: How to push fine-tuned model to Hugging Face hub

Use the huggingface_hub Python library to push your fine-tuned model to the Hugging Face hub by logging in with your token and calling push_to_hub() on your model or repository. This process uploads model files, configuration, and tokenizer for easy sharing and deployment.

Quick answer

Use the huggingface_hub Python library to push your fine-tuned model to the Hugging Face hub by logging in with your token and calling push_to_hub() on your model or repository. This process uploads model files, configuration, and tokenizer for easy sharing and deployment.

PREREQUISITES

Python 3.8+
pip install huggingface_hub transformers
Hugging Face account with an access token

Setup

Install the huggingface_hub and transformers libraries and set your Hugging Face access token as an environment variable for authentication.

bash

pip install huggingface_hub transformers

Step by step

This example shows how to push a fine-tuned transformers model to the Hugging Face hub using the push_to_hub() method.

python

import os
from transformers import AutoModelForSequenceClassification, AutoTokenizer

# Load your fine-tuned model and tokenizer
model_name_or_path = "./fine_tuned_model"
model = AutoModelForSequenceClassification.from_pretrained(model_name_or_path)
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)

# Set your HF token in environment variable HF_TOKEN
hf_token = os.environ["HF_TOKEN"]

# Define your repo name on Hugging Face hub
repo_name = "your-username/fine-tuned-model"

# Push model and tokenizer to the hub
model.push_to_hub(repo_name, use_auth_token=hf_token)
tokenizer.push_to_hub(repo_name, use_auth_token=hf_token)

print(f"Model and tokenizer pushed to https://huggingface.co/{repo_name}")

output

Model and tokenizer pushed to https://huggingface.co/your-username/fine-tuned-model

Common variations

Use huggingface_hub.Repository class to manage repo files and commits manually.
Push other model types like AutoModelForCausalLM similarly.
Use use_auth_token=True if your token is set in ~/.huggingface/token.

python

from huggingface_hub import Repository

repo = Repository(local_dir="./fine_tuned_model", clone_from=repo_name, use_auth_token=hf_token)
repo.push_to_hub(commit_message="Add fine-tuned model files")

Troubleshooting

If you get authentication errors, verify your HF_TOKEN environment variable is set correctly.
Ensure your repo name includes your username and is unique on Hugging Face hub.
Check your internet connection and firewall settings if upload fails.

✅

Key Takeaways

Use the huggingface_hub Python library's push_to_hub() method to upload fine-tuned models.
Set your Hugging Face access token in an environment variable for secure authentication.
You can push both model weights and tokenizer files to the Hugging Face hub.
The Repository class allows manual control over repo commits and file management.
Verify repo naming and token permissions to avoid common push errors.

Verified 2026-04 · AutoModelForSequenceClassification, AutoTokenizer

Verify ↗