input: "BoxPredictor_4/ClassPredictor/weights_quant/max/read"
  attr {
    key: "narrow_range"
    value {
      b: true
    }
  }
  attr {
    key: "num_bits"
    value {
      i: 8
    }
  }
}
node {
  name: "BoxPredictor_4/ClassPredictor/Conv2D"
  op: "Conv2D"
  input: "input"

  if (!IsLegalQuantSpecs(func)) {
    return true;
  }

  OpBuilder builder(func);
  bool is_signed = quant_specs_.IsSignedInferenceType();
  IntegerAttr num_bits =
      builder.getI32IntegerAttr(quant_specs_.GetQuantizationTypeWidth());
  BoolAttr narrow_range = builder.getBoolAttr(false);

  auto add_quantize_op = [&](Location loc, Type input_type, Block* block,

	format.Erasure.Version = formatErasureVersionV3
	format.Erasure.DistributionAlgo = formatErasureVersionV3DistributionAlgoV3
	format.Erasure.Sets = make([][]string, numSets)

	for i := 0; i < numSets; i++ {
		format.Erasure.Sets[i] = make([]string, setLen)
		for j := 0; j < setLen; j++ {
			format.Erasure.Sets[i][j] = mustGetUUID()
		}
	}
	return format
}

  if (!IsLegalQuantSpecs(func)) {
    return true;
  }

  OpBuilder builder(func);
  bool is_signed = quant_specs_.IsSignedInferenceType();
  IntegerAttr num_bits =
      builder.getI32IntegerAttr(quant_specs_.GetQuantizationTypeWidth());
  BoolAttr narrow_range = builder.getBoolAttr(false);

  auto add_quantize_op = [&](Location loc, Type input_type, Block* block,

    return op.emitOpError("range is invalid: [" + Twine(std::to_string(rmin)) +
                          "," + Twine(std::to_string(rmax)) + "]");
  }
  int64_t num_bits = op.getNumBits();
  if (num_bits < 2 || num_bits > 16) {
    return op.emitOpError(
        "requires num_bits to be between 2 and 16, inclusive");
  }
  return success();
}

//===----------------------------------------------------------------------===//

					numValues[i] = v
					numUnit[i] = unit
				}
				numLabels[k] = append(numLabels[k], numValues...)
				numUnits[k] = append(numUnits[k], numUnit...)
			}
		}
		s.NumLabel = numLabels
		s.NumUnit = numUnits
	}

	// Remove label marking samples from the base profiles, so it does not appear
	// as a nodelet in the graph view.
	prof.RemoveLabel("pprof::base")

}

// numLabelsToString returns a string representation of a map
// representing numeric labels.
func numLabelsToString(numLabels map[string][]int64, numUnits map[string][]string) string {
	ls := []string{}
	for k, v := range numLabels {
		units := numUnits[k]
		var labelString string
		if len(units) == len(v) {
			values := make([]string, len(v))
			for i, vv := range v {

  %0 = "tf.FakeQuantWithMinMaxVars"(%arg0, %arg1, %arg2) {min = 0.0 : f32, max = 1.0 : f32, num_bits = 5, narrow_range = true} : (tensor<8x8x8x8xf32>, tensor<f32>, tensor<f32>) -> tensor<8x8x8x8xf32>
  func.return %0 : tensor<8x8x8x8xf32>

  // CHECK-LABEL: fakeQuantVarsTrue
  // CHECK: "tf.FakeQuantWithMinMaxVars"(%arg0, %arg1, %arg2) <{narrow_range = true, num_bits = 5 : i64}> {max = 1.000000e+00 : f32, min = 0.000000e+00 : f32}
}

        attr, it->second, /*symmetric=*/true, /*num_bits=*/8, is_signed_,
        /*narrow_range=*/true, legacy_float_scale_);
  } else {
    // per-tensor quantization weight
    final_type = GetUniformQuantizedTypeForWeight(
        attr, /*symmetric=*/is_weight && is_signed_,
        /*num_bits=*/8, is_signed_,
        /*narrow_range=*/is_weight, legacy_float_scale_);
  }

            max=ops.convert_to_tensor(self._max[0]),
            num_bits=8,
            narrow_range=False,
        )
        y = array_ops.fake_quant_with_min_max_vars(
            y,
            min=ops.convert_to_tensor(self._min[1]),
            max=ops.convert_to_tensor(self._max[1]),
            num_bits=8,
            narrow_range=False,
        )

        out = math_ops.matmul(x, y)