Hi, I'm trying to deploy my AI model to gap8 using gapflow, but I always encounter truncated data during quantisation, e.g. all the outputs of a certain layer will be 127 or -128. Therefore, I want to know if nntool can help me directly perform mixed-precision quantisation.
Hi, I'm trying to deploy my AI model to gap8 using gapflow, but I always encounter truncated data during quantisation, e.g. all the outputs of a certain layer will be 127 or -128. Therefore, I want to know if nntool can help me directly perform mixed-precision quantisation.