Introducing Quanto: A New Quantization Backend for PyTorch in Optimum

Hugging Face has announced Quanto, a new quantization backend for PyTorch integrated into the Optimum library. Quanto is designed to simplify model quantization, enabling developers to reduce model size and accelerate inference with minimal accuracy loss. It supports both dynamic and static quantization, and works seamlessly with popular transformer models. This addition expands Optimum's hardware optimization capabilities, making it easier to deploy PyTorch models on resource-constrained devices.