Part Number: TDA4VE-Q1
when the model is converted in the sdk, if both numParamBits and numFeatureBits in the import config are set to 8 bits, the model's inference results will differ significantly from those of the pc. howwever, if they are set to 16…