// fixed upstream).
func.func @broadcast_add(%arg0: tensor<1xi32>, %arg1: tensor<1x2xi32>) -> tensor<1x2xi32> {
  // CHECK-NEXT: %[[LHS_BCAST:.+]] = "mhlo.broadcast_in_dim"(%arg0) <{broadcast_dimensions = dense<1> : tensor<1xi64>}>
  // CHECK-NEXT: mhlo.add %[[LHS_BCAST]], %arg1
  %0 = "tf.AddV2"(%arg0, %arg1) : (tensor<1xi32>, tensor<1x2xi32>) -> tensor<1x2xi32>
  func.return %0: tensor<1x2xi32>
}

// CHECK-LABEL: func @broadcast_multi_dim_add

  %split_dim_2 = arith.constant dense<1> : tensor<1xi32>
  %4, %5 = "tfl.split"(%split_dim_2, %arg0) {num_splits = 2 : i32} : (tensor<1xi32>, tensor<16x4xf32>) -> (tensor<16x2xf32>, tensor<16x2xf32>)
  %6:2 = "tfl.split"(%split_dim_2, %arg0) {num_splits = 2 : i32} : (tensor<1xi32>, tensor<16x4xf32>) -> (tensor<16x2xf32>, tensor<16x?xf32>)

      %%arg0 = "tf.Const"() {value = dense<-3.0> : tensor<1x4x2xf32>} : () -> tensor<1x4x2xf32>
      %%arg1 = "tf.Const"() {value = dense<-3.0> : tensor<1x2x4xf32>} : () -> tensor<1x2x4xf32>

      %%1 = "tf.%s"(%%arg0, %%arg1) {T = f32, adj_x = false, adj_y = false, grad_x = false, grad_y = false, device = ""} : (tensor<1x4x2xf32>, tensor<1x2x4xf32>) -> tensor<1x4x4xf32>

      func.return %%1 : tensor<1x4x4xf32>

    %1 = "tf.Const"() {value = dense<1.000000e+00> : tensor<f32>} : () -> tensor<f32>
    %2 = "tf.Const"() {value = dense<-1> : tensor<i32>} : () -> tensor<i32>
    %3 = "tf.Const"() {value = dense<[[0, 1]]> : tensor<1x2xi32>} : () -> tensor<1x2xi32>
    %4 = "tf.Const"() {value = dense<> : tensor<0xi32>} : () -> tensor<0xi32>
    %5 = "tf.Const"() {value = dense<2.500000e-01> : tensor<f32>} : () -> tensor<f32>

}

// -----

// CHECK-LABEL: func @select_batch_static_r1
func.func @select_batch_static_r1(%arg0: tensor<i1>, %arg1: tensor<2x6x8xi32>, %arg2: tensor<2x6x8xi32>) -> tensor<2x6x8xi32> {
  // CHECK: mhlo.select %arg0, %arg1, %arg2
  %0 = "tf.Select"(%arg0, %arg1, %arg2) : (tensor<i1>, tensor<2x6x8xi32>, tensor<2x6x8xi32>) -> tensor<2x6x8xi32>
  func.return %0: tensor<2x6x8xi32>
}

// -----

  %group_assignment = "tf.Const"() { value = dense<[[0, 1]]> : tensor<1x2xi32> } : () -> tensor<1x2xi32>
  %instance_key = "tf.Const"() { value = dense<3> : tensor<i32> } : () -> tensor<i32>
  %group_size, %group_key = "tf.CollectiveAssignGroupV2"(%group_assignment, %rank, %key_base) {} : (tensor<1x2xi32>, tensor<i32>, tensor<i32>) -> (tensor<i32>, tensor<i32>)
  // CHECK-NOT: "tf.CollectiveAssignGroupV2"

}

func.func @einsum_reshapetail(%arg0: tensor<3x4x5xf32>, %arg1: tensor<5x6x2xf32>) -> tensor<3x4x6x2xf32> {
  %0 = "tf.Einsum"(%arg0, %arg1) {T = "tfdtype$DT_FLOAT", equation = "bfd,dnh->bfnh"}: (tensor<3x4x5xf32>, tensor<5x6x2xf32>) -> tensor<3x4x6x2xf32>
  func.return %0 : tensor<3x4x6x2xf32>
  // CHECK-LABEL: einsum_reshapetail

func.func @expandDims(%arg0: tensor<2x2xf32>, %arg1: tensor<i32>) -> tensor<1x2x2xf32> {
  %0 = "tf.ExpandDims"(%arg0, %arg1) : (tensor<2x2xf32>, tensor<i32>) -> tensor<1x2x2xf32>
  func.return %0 : tensor<1x2x2xf32>

// CHECK-LABEL:expandDims
// CHECK:  "tfl.expand_dims"(%arg0, %arg1) : (tensor<2x2xf32>, tensor<i32>) -> tensor<1x2x2xf32>
}

func.func @squeezeDefault(%arg0: tensor<1x2x2xf32>) -> tensor<2x2xf32> {

// Examples:
//   * If `xla_gather_op_output_type` == tensor<*xf32>, then it returns:
//     tensor<*xf32>.
//   * If `xla_gather_op_output_type` == tensor<3x5xi32> and `collapsed_dims` ==
//     {0}, then it returns: tensor<1x3x5xi32>.
//   * If `xla_gather_op_output_type` == tensor<3x5xf32> and `collapsed_dims` ==
//     {1, 3}, then it returns: tensor<3x1x5x1xf32>.
Type GetSliceOpOutputType(Type xla_gather_op_output_type,

func.func @reshape_vector_shape(tensor<4x4x4xf32>) -> tensor<16x4xf32> {
^bb0(%arg0: tensor<4x4x4xf32>) :
  %shape0 = arith.constant dense<[[16, 4]]> : tensor<1x2xi32>
  // expected-error @+1 {{'tfl.reshape' op requires 'shape' to be rank 1, but got 2}}
  %1 = "tfl.reshape"(%arg0, %shape0) : (tensor<4x4x4xf32>, tensor<1x2xi32>) -> tensor<16x4xf32>
  func.return %1 : tensor<16x4xf32>
}

// -----