}
	// iterate through the above set of inputs and upload the object.
	for i, input := range putObjectInputs {
		// uploading the object.
		_, err := obj.PutObject(context.Background(), input.bucketName, input.objectName, mustGetPutObjReader(t, bytes.NewReader(input.textData), input.contentLength, input.metaData[""], ""), ObjectOptions{UserDefined: input.metaData})
		// if object upload fails stop the test.
		if err != nil {

  //   invariant, this does not hold as shapes can differ.
  // * All three pairs using cond inputs, body inputs and results as operand is
  //   a common source for all three.
  // * Body result and cond inputs to call the cond function for the subsequent
  //   iterations. Similarly, Body result should be compatible with body inputs
  //   and op results.
  //

    }
    for (auto edge : n->in_edges()) {
      inputs[index] =
          NodeDefBuilder::NodeOut{edge->src()->name(), edge->src_output(),
                                  edge->src()->output_type(edge->src_output())};
    }
  }
  send_from_host_builder.Input(inputs);
  send_from_host_builder.Input(key_placeholder->name(), 0, DT_STRING);

    # Create and save a simple model that involves a hash table.
    inputs, outputs = self._create_and_save_vocab_table_lookup_qat_model_tf1(
        self._input_saved_model_path, tags, signature_def_key
    )

    # Make sure that the desired input key and output key is present.
    self.assertIn('input_vocabs', inputs.keys())
    self.assertIn('lookup', outputs.keys())

      return op.emitOpError(
          "has mismatched quantized axes of input and output");
    }
  }

  return success();
}

static void BuildTransposeOp(OpBuilder* builder, OperationState& result,
                             Value input, Value perm) {
  // Output size is only known if input is ranked and perm is a constant.
  auto input_type = input.getType().cast<TensorType>();
  DenseIntElementsAttr perm_const;

    func.return %1 : tensor<2xi32>
  }

  //  Tests that all inputs to a function are passed and all outputs are returned for an outside compilation cluster
  //  with dynamic-shaped inputs/outputs. This test case tests that if another outside compiled returns
  //  only statically-shaped outputs, the value consumed by an outside compiled op with some dynamically shaped inputs is
  //  sent back to the host.

  if (max && !IsOfRankedFloatTensorType(max, 1))
    return op.emitOpError("requires max to be a 1d float tensor");

  Value inputs = op.getInputs();
  if (!HasRankAtLeast(inputs, 1))
    return op.emitError("requires inputs to be at least 1d float tensor");

  int64_t num_bits = op.getNumBits();
  if (num_bits < 2 || num_bits > 16) {
    return op.emitOpError(

                               " input ", i, ", expected control input ",
                               a.input(i), " got ", b.input(i), " expected:\n",
                               a.DebugString(), "\ngot:\n", b.DebugString());
        }
        return false;
      }
      control_input_a.insert(a.input(i));
      control_input_b.insert(b.input(i));
    } else if (a.input(i) != b.input(i)) {
      if (diff) {

      SmallVector<ValuePort, 4> inputs;
      auto res = ComputeInputsRequiredForOutput(front, &inputs);
      if (failed(res)) {
        // Abort if unable to find which required inputs need to be computed.
        worklist.clear();
        break;
      }

      if (!inputs.empty()) {
        // Enqueue required computation followed by its required operands in

    func.return %1, %3 : tensor<*xi32>, tensor<*xi1>
  }
}

// -----

// Tests inputs are correctly split and fed into TPU computation for tiled input
// sharding.

// The following OpSharding is used for TPU computation inputs in below test:
// Proto debug string:
//  input 0
//   type: OTHER
//   tile_assignment_dimensions: 1
//   tile_assignment_dimensions: 2