tf.fake_quant_with_min_max_vars_per_channel
tf.fake_quant_with_min_max_vars_per_channel
tf.fake_quant_with_min_max_vars_per_channel
fake_quant_with_min_max_vars_per_channel( inputs, min, max, num_bits=None, name=None )
Defined in tensorflow/python/ops/gen_array_ops.py
.
See the guide: Tensor Transformations > Fake quantization
Fake-quantize the 'inputs' tensor of type float and one of the shapes: [d]
,
[b, d]
[b, h, w, d]
via per-channel floats min
and max
of shape [d]
to 'outputs' tensor of same shape as inputs
.
[min; max] is the clamping range for the 'inputs' data in the corresponding depth channel. Op divides this range into 255 steps (total of 256 values), then replaces each 'inputs' value with the closest of the quantized step values. 'num_bits' is the bitwidth of the quantization; between 2 and 8, inclusive.
This operation has a gradient and thus allows for training min
and max
values.
Args:
-
inputs
: ATensor
of typefloat32
. -
min
: ATensor
of typefloat32
. -
max
: ATensor
of typefloat32
. -
num_bits
: An optionalint
. Defaults to8
. -
name
: A name for the operation (optional).
Returns:
A Tensor
of type float32
.
© 2017 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 3.0.
Code samples licensed under the Apache 2.0 License.
https://www.tensorflow.org/api_docs/python/tf/fake_quant_with_min_max_vars_per_channel