Searched refs:quantization_bias (Results 1 – 1 of 1) sorted by relevance
23 float DequantizeValue(int num_sparse_features, int quantization_bias, in DequantizeValue() argument25 return 1.0 / num_sparse_features * (value - quantization_bias) * multiplier; in DequantizeValue()44 const int quantization_bias = 1 << (quantization_bits - 1); in DequantizeAddNBit() local58 dest[i] += DequantizeValue(num_sparse_features, quantization_bias, in DequantizeAddNBit()