Migrate datasets from GitHub to Hugging Face Hub
Upload files to a Hugging Face repository
Annotation Tool
Generate a Parquet file for dataset validation
Data annotation for Sparky
Validate JSONL format for fine-tuning
Support by Parquet, CSV, Jsonl, XLS
List of French datasets not referenced on the Hub
Manage and label your datasets
Create datasets with FAQs and SFT prompts
Convert a model to Safetensors and open a PR
Build datasets using natural language
Search narrators and view network connections
The Github To Huggingface Dataset Migration Tool is a utility designed to simplify the process of migrating datasets from GitHub repositories to the Hugging Face Hub. It enables users to seamlessly transfer their dataset files, maintaining the integrity and structure of the data. This tool is particularly useful for researchers and data scientists who want to leverage Hugging Face's ecosystem for dataset sharing, versioning, and collaboration.
• Support for GitHub Repositories: Migrate datasets stored in public or private GitHub repositories.
• Dataset Versioning: Preserve dataset versions during the migration process.
• Large Dataset Handling: Efficiently manage and transfer large-scale datasets.
• Data Format Compatibility: Supports various data formats commonly used in machine learning workflows.
• Integration with Hugging Face Libraries: Compatibility with Hugging Face's ecosystem, including datasets and transformers libraries.
• Repository Structure Mapping: Maintain the original folder and file structure of the dataset during migration.
• Commit History Preservation: Optionally retain commit history and metadata from GitHub.
• Validation and Verification: Ensure data integrity through pre- and post-migration checks.
What types of GitHub repositories are supported?
The tool supports both public and private GitHub repositories. For private repositories, ensure you have the necessary authentication credentials.
Can I migrate datasets that are split across multiple repositories?
Yes, the tool allows you to specify multiple repositories or subdirectories within a repository to migrate as a single dataset.
How are dataset versions handled during migration?
The tool preserves the version history by creating separate versions on Hugging Face Hub. You can also choose to create a new version or overwrite an existing one.