/tensorflow-refmodel

Based on the trained model and the customized structure, 8bit quantization weight is generated for the chip, and quantization accuracy is output

Primary LanguageJupyter Notebook

Watchers