This tutorial demonstrates how to apply 'INT8' quantization to the Big Transfer Image Classification model. Here we demonstrate fine-tuning the model, OpenVINO optimization and followed by INT8 quantization processing with NNCF.
This tutorial consists of the following steps:
- Prepare Dataset.
- Plotting data samples.
- Model fine-tuning.
- Perform model optimization (IR) step.
- Compute model accuracy of the TF model.
- Compute model accuracy of the optimized model.
- Run nncf.Quantize for getting an Optimized model.
- Compute model accuracy of the quantized model.
- Compare model accuracy of the optimized and quantized models.
- Compare inference results on one picture
If you have not installed all required dependencies, follow the Installation Guide