%cst_0 = arith.constant dense<[2, -1]> : tensor<2xi64>
  %0 = "tf.FakeQuantWithMinMaxVars"(%arg0, %arg1, %arg2) {num_bits = 3, narrow_range = false} : (tensor<1x2xf32>, tensor<f32>, tensor<f32>) -> tensor<1x2xf32>
  %1 = "tf.Reshape"(%0, %cst_0) : (tensor<1x2xf32>, tensor<2xi64>) -> tensor<2x1xf32>
  %2 = "tf.Reshape"(%0, %cst_0) : (tensor<1x2xf32>, tensor<2xi64>) -> tensor<2x1xf32>
  func.return %1, %2 : tensor<2x1xf32>, tensor<2x1xf32>

    %range_nume = "tf.Sub"(%r_max_f64, %r_min_f64) : (tensor<1xf64>, tensor<1xf64>) -> tensor<1xf64>
    %range_deno = "tf.Sub"(%i8_max_f64, %i8_min_f64) : (tensor<f64>, tensor<f64>) -> tensor<f64>

    %scale_f64 = "tf.Div"(%range_nume, %range_deno) : (tensor<1xf64>, tensor<f64>) -> tensor<1xf64>
    %scale = "tf.Cast"(%scale_f64) : (tensor<1xf64>) -> tensor<1xf32>

    // Add comparison with minimum if needed

        %1 = "tf.OpB"() {_XlaSharding = "\08\03\1A\02\02\01\22\02\00\02"} : () -> tensor<2x2xi64>
        // expected-error @+1 {{should have the same sharding}}
        %2 = "tf.OpC"(%0, %1) {_xla_map_outside_compilation = "0", _xla_outside_compilation = "from_launch"} : (tensor<2x2xi64>, tensor<2x2xi64>) -> tensor<2x2xi64>
        "tf.OpD"(%2) : (tensor<2x2xi64>) -> ()
        tf_device.return

  %1 = "tfl.shape"(%conv_s) : (tensor<?x26x26x26x16xf32>) -> tensor<5xi64>
  %2 = "tfl.broadcast_args"(%1, %cst) : (tensor<5xi64>, tensor<1xi64>) -> tensor<5xi64>
  %broad1 = "tfl.broadcast_to"(%conv_s, %2) : (tensor<?x26x26x26x16xf32>, tensor<5xi64>) -> tensor<?x26x26x26x16xf32>
  %broad2 = "tfl.broadcast_to"(%b, %2) : (tensor<16xf32>, tensor<5xi64>) -> tensor<?x26x26x26x16xf32>

  // CHECK-LABEL: func @init_all_tables_with_explicit_vocab_size
  // CHECK-DAG: [[CST:%.*]]  = arith.constant dense<["apple", "banana"]> : tensor<2x!tf_type.string>
  // CHECK-DAG: [[CST_0:%.*]]  = arith.constant dense<[0, 1]> : tensor<2xi64>
  // CHECK: [[VAL:%.*]] = "tf.HashTableV2"()
  // CHECK: "tf.LookupTableImportV2"([[VAL]], [[CST]], [[CST_0]])

    %24 = "tf.Softmax"(%18) : (tensor<2x1001xf32>) -> tensor<2x1001xf32>
    %25 = "tf.ArgMax"(%24, %2) : (tensor<2x1001xf32>, tensor<i32>) -> tensor<2xi64>
    %26 = "tf.Cast"(%25) {Truncate = false} : (tensor<2xi64>) -> tensor<2xf32>
    %27 = "tf.Equal"(%14, %26) {incompatible_shape_error = true} : (tensor<2xf32>, tensor<2xf32>) -> tensor<2xi1>

func.func @testBroadcastGradientArgI64() -> (tensor<2xi64>, tensor<0xi64>) {
  %s0 = "tf.Const"() {value = dense<> : tensor<0xi64>} : () -> tensor<0xi64>
  %s1 = "tf.Const"() {value = dense<[2, 4]> : tensor<2xi64>} : () -> tensor<2xi64>
  %r0, %r1 = "tf.BroadcastGradientArgs"(%s0, %s1) {} : (tensor<0xi64>, tensor<2xi64>) -> (tensor<2xi64>, tensor<0xi64>)

  %2 = "tf.Shape"(%0) {device = "/CPU:0"} : (tensor<*xi32>) -> tensor<?xi64>
  %dyn_2 = "tf.SetStaticDimensionBounds" (%2, %arg4) :(tensor<?xi64>, tensor<?xi64>) -> tensor<?xi64>
  %compilation_status, %program = "tf._TPUCompileMlir"(%1, %2) {device = "/CPU:0", metadata = "metadata", mlir_module = "mlir_module"} : (tensor<?xi64>, tensor<?xi64>) -> (tensor<!tf_type.string>, tensor<3x!tf_type.string>)

    func.func @main() -> () {
      %cst0 = "tf.Const"(){ value = dense<0> : tensor<3x5xi1>} : () -> tensor<3x5xi1>
      %0 = "tf.Where"(%cst0) : (tensor<3x5xi1>) -> tensor<?x2xi64>
      func.return
    }
  })";

  CellReader<int64_t> dynamism_counter(kDynamismOpCounterStreamzName);
  CellReader<int64_t> dynamism_function_counter(
      kDynamismFunctionCounterStreamzName);

  %7 = "mhlo.broadcast_in_dim"(%3) <{broadcast_dimensions = dense<3> : tensor<1xi64>}> : (tensor<5xf32>) -> tensor<32x28x28x5xf32>
  %8 = mhlo.add %6, %7 : tensor<32x28x28x5xf32>
  %9 = mhlo.maximum %8, %1 : tensor<32x28x28x5xf32>
  %10 = "mhlo.reshape"(%9) : (tensor<32x28x28x5xf32>) -> tensor<32x3920xf32>