Searched refs:input_product_scale (Results 1 – 1 of 1) sorted by relevance
112 const double input_product_scale = input->params.scale * filter->params.scale; in GetQuantizedConvolutionMultipler() local118 TF_LITE_ENSURE(context, std::abs(input_product_scale - bias_scale) <= in GetQuantizedConvolutionMultipler()119 1e-6 * std::min(input_product_scale, bias_scale)); in GetQuantizedConvolutionMultipler()120 TF_LITE_ENSURE(context, input_product_scale >= 0); in GetQuantizedConvolutionMultipler()122 *multiplier = input_product_scale / output_scale; in GetQuantizedConvolutionMultipler()