Home
last modified time | relevance | path

Searched refs:quantized_value (Results 1 – 8 of 8) sorted by relevance

/external/tensorflow/tensorflow/examples/speech_commands/
Dwav_to_features.py103 quantized_value = int(
106 if quantized_value < 0:
107 quantized_value = 0
108 if quantized_value > 255:
109 quantized_value = 255
112 f.write('%d, ' % (quantized_value))
/external/tensorflow/tensorflow/lite/tools/optimize/
Dquantization_utils.cc152 const int32_t quantized_value = in SymmetricPerChannelQuantizeValues() local
156 std::max<int8_t>(kMinQuantizedValue, quantized_value)); in SymmetricPerChannelQuantizeValues()
Dsubgraph_quantizer.cc180 const int32_t quantized_value = static_cast<int32_t>( in SymmetricPerChannelBiasQuantize() local
183 std::min(kScale, std::max(-kScale, quantized_value)); in SymmetricPerChannelBiasQuantize()
/external/gemmlowp/doc/
Dquantization.md40 real_value = A * quantized_value + B (1)
46 real_value = C * (quantized_value + D) (2)
80 In equation (1), plugging `real_value = 0` and `quantized_value = zero_point`,
98 `quantized_value = zero_point`, we get:
116 real_value = scale * (quantized_value - zero_point) (3)
124 possible `quantized_value`'s, and equation (3) maps it to a `real_value` of
/external/tensorflow/tensorflow/lite/kernels/internal/reference/
Dportable_tensor_utils.cc61 const int32_t quantized_value = in PortableSymmetricQuantizeFloats() local
64 quantized_values[i] = std::min(kScale, std::max(-kScale, quantized_value)); in PortableSymmetricQuantizeFloats()
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/
Dneon_tensor_utils.cc953 const int32 quantized_value = in NeonSymmetricQuantizeFloats() local
955 quantized_values[i] = std::min(kScale, std::max(-kScale, quantized_value)); in NeonSymmetricQuantizeFloats()
/external/tensorflow/tensorflow/lite/g3doc/guide/
Dios.md212 float_value = (quantized_value - zero_point) * scale
/external/tensorflow/tensorflow/lite/g3doc/models/image_classification/
Dios.md212 float_value = (quantized_value - zero_point) * scale