We loaded Model 1 into STM32Cube.ai and selected compression by a factor of 4. As result a limited number of floating point weights are stored in a look-up ...
確定! 回上一頁