Convert a model to Safetensors and open a PR
Convert and PR models to Safetensors
Browse and search datasets
Speech Corpus Creation Tool
Evaluate evaluators in Grounded Question Answering
Create a large, deduplicated dataset for LLM pre-training
Save user inputs to datasets on Hugging Face
Download datasets from a URL
Search for Hugging Face Hub models
Create datasets with FAQs and SFT prompts
Upload files to a Hugging Face repository
Perform OSINT analysis, fetch URL titles, fine-tune models
Convert to Safetensors is a practical tool designed to convert AI models to the Safetensors format, a modern, efficient alternative to older formats like PyTorch's .pt files. Its primary purpose is to simplify model conversion and ** FACILITATE the creation of pull requests (PRs)** for integrating these models into larger projects or libraries. By using this tool, users can ensure their models are compatible with the latest standards in AI model sharing.
pip install convert-to-safetensors
to install the package.convert-to-safetensors --input-path /path/to/model --output-path /path/to/output
.What is the Safetensors format?
Safetensors is a secure and efficient format for storing and sharing AI models, designed to replace older formats like PyTorch's .pt files. It offers better security and compatibility across different frameworks.
Can I use Convert to Safetensors for non-PyTorch models?
Yes, Convert to Safetensors supports models from various frameworks, including TensorFlow and others, making it a versatile tool for AI practitioners.
Where are the converted models stored?
The converted models are stored in the output directory you specify during the conversion process. Ensure you have write permissions for the target directory.