# RUN: tf_tfl_translate -tf-input-arrays=input -tf-input-shapes=4 -tf-input-data-types=DT_INT32 -tf-output-arrays=output %s -o - --output-mlir | FileCheck %s

node {
  name: "default"
  op: "Const"
  attr {
    key: "dtype"
    value {
      type: DT_INT32
    }
  }
  attr {
    key: "value"
    value {
      tensor {
        dtype: DT_INT32
        tensor_shape {
        }
        int_val: 0
      }
    }
  }
}

        file("input.txt").createNewFile()
        file("input-nested.txt").createNewFile()
        file("input1.txt").createNewFile()
        file("input2.txt").createNewFile()
        file("inputs").createDir()
        file("inputs/inputA.txt").createNewFile()
        file("inputs/inputB.txt").createNewFile()
        file("inputs1").createDir()
        file("inputs2").createDir()

        buildFile """

# Verify that we match correctly the input / output when they are scalar.

# CHECK-LABEL: func @main
# CHECK-SAME:  (%{{[a-z0-9]+}}: tensor<f32> {tf.device = "/device:CPU:0"})
# CHECK-SAME:  control_outputs = ""
# CHECK-SAME:  inputs = "input"
# CHECK-SAME:  outputs = "out:1,out"

      else:
        # Op has been registered, then we don't need to generate register code.
        return

    # Validates the function inputs match what are in the decorator.
    inputs = all_dec_args.get('inputs', [])
    attrs = all_dec_args.get('attrs', [])
    expected_args = [arg.split(':')[0] for arg in inputs + attrs]
    all_func_args = self.visit(node.args)

    if len(expected_args) != len(all_func_args):
      raise KeyError(

is used instead.

Operands are replicated inputs and packed inputs.

replicated_inputs: each group of `n` inputs corresponds to an input for a single
individual replica and is mapped to a single region argument. Inside one group
the operands are matching in order the `devices` attribute. Each replicated
input must have compatible shapes and types.
packed_inputs: each input corresponds to an input broadcasted across all

               op_name,
               inputs=None,
               attrs=None,
               derived_attrs=None,
               outputs=None):
    self._op_name = op_name
    self._inputs = inputs
    self._attrs = attrs
    self._derived_attrs = derived_attrs
    self._outputs = outputs

  def __call__(self, compose_fn):
    # TODO(fengliuai): more sanity check of the input function and make sure

	// swap in and out to exercise copy-up and copy-down
	input8, output8 = output8, input8
	input16, output16 = output16, input16
	input32, output32 = output32, input32
	input64, output64 = output64, input64
	in := 0
	out := 13
	for i := range input8 {
		input8[i] = u8(in)
		output8[i] = u8(out)
		outputS[i] = u8(out)
		input16[i] = u16(in)
		output16[i] = u16(out)
		input32[i] = u32(in)
		output32[i] = u32(out)

inline SignatureDefT::SignatureDefT(const SignatureDefT &o)
      : signature_key(o.signature_key),
        subgraph_index(o.subgraph_index) {
  inputs.reserve(o.inputs.size());
  for (const auto &inputs_ : o.inputs) { inputs.emplace_back((inputs_) ? new tflite::TensorMapT(*inputs_) : nullptr); }
  outputs.reserve(o.outputs.size());

Robert Griesemer <******@****.***> 1408667113 -0700

Robert Griesemer <******@****.***> 1408667113 -0700