tf.quantization.fake_quant_with_min_max_vars_per_channel_gradient

Aliases:

  • tf.fake_quant_with_min_max_vars_per_channel_gradient
  • tf.quantization.fake_quant_with_min_max_vars_per_channel_gradient
tf.quantization.fake_quant_with_min_max_vars_per_channel_gradient(
    gradients,
    inputs,
    min,
    max,
    num_bits=8,
    narrow_range=False,
    name=None
)

Defined in generated file: tensorflow/python/ops/gen_array_ops.py.

Compute gradients for a FakeQuantWithMinMaxVarsPerChannel operation.

Args:

  • gradients: A Tensor of type float32. Backpropagated gradients above the FakeQuantWithMinMaxVars operation, shape one of: [d], [b, d], [b, h, w, d].
  • inputs: A Tensor of type float32. Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape same as gradients. min, max: Quantization interval, floats of shape [d].
  • min: A Tensor of type float32.
  • max: A Tensor of type float32.
  • num_bits: An optional int. Defaults to 8. The bitwidth of the quantization; between 2 and 16, inclusive.
  • narrow_range: An optional bool. Defaults to False. Whether to quantize into 2^num_bits - 1 distinct values.
  • name: A name for the operation (optional).

Returns:

A tuple of Tensor objects (backprops_wrt_input, backprop_wrt_min, backprop_wrt_max).

  • backprops_wrt_input: A Tensor of type float32.
  • backprop_wrt_min: A Tensor of type float32.
  • backprop_wrt_max: A Tensor of type float32.