AIMET ONNX Quantization APIs

AIMET Quantization for ONNX Models provides the following functionality.

• Quantization Simulation API: Allows ability to simulate inference on quantized hardware

• Cross-Layer Equalization API: Post-training quantization technique to equalize layer parameters

• Adaround API: Post-training quantization technique to optimize rounding of weight tensors

• AutoQuant API: Unified API that integrates the post-training quantization techniques provided by AIMET