Exporting Quantized Models

Quark torch not only supports our own torch export format Quark format (Json-Pth), but also support exporting in popular formats requested by downstream tools, including ONNX, format for Hugging Face & vLLM (HF format), and GGUF.

• ONNX format
• Hugging Face format (safetensors)
• GGUF format
• Quark format