Hugging Face Node

Interact with Hugging Face Hub: run inference, create models/spaces/datasets, and upload/download files.

Features

Inference

• Run Inference: Execute inference on any Hugging Face model
  • Supports multiple task types (text generation, classification, QA, etc.)
  • Auto-detect task type or specify manually
  • Custom parameters support
  • Wait for model loading option

Repository Management

• Create Model: Create new model repositories
• Create Space: Create new Spaces with different SDKs (Gradio, Streamlit, Docker, Static)
• Create Dataset: Create new dataset repositories

File Operations

• Upload Model: Upload files to model repositories
• Upload Dataset: Upload files to dataset repositories
• Download Model: Download files from model repositories
• Download Dataset: Download files from dataset repositories

Authentication

All actions require a Hugging Face access token:

1. Visit Hugging Face Settings
2. Create a new token with appropriate permissions:
   • Read token: For inference and downloads
   • Write token: For creating repositories and uploading files

Usage Examples

Run Text Generation Inference

{
  "model": "gpt2",
  "task": "text-generation",
  "inputs": "The future of AI is",
  "parameters": {
    "max_length": 50,
    "temperature": 0.7
  }
}

Run Classification

{
  "model": "distilbert-base-uncased-finetuned-sst-2-english",
  "task": "text-classification",
  "inputs": "I love this product!"
}

Create Model Repository

{
  "modelId": "username/my-model",
  "private": false
}

Upload Model File

{
  "modelId": "username/my-model",
  "filePath": "/path/to/model.bin",
  "pathInRepo": "pytorch_model.bin",
  "commitMessage": "Upload model weights"
}

API References

• Hugging Face Inference API
• Hugging Face Hub API
• Hugging Face Documentation

Notes

• For large file uploads/downloads, consider using the official huggingface_hub Python library
• Some operations may require write permissions on your token
• Model inference may take time on first request (cold start)