%output_0, %min_1, %max_2, %histogram_3 = "tf.CustomAggregator"(%0) <{calibration_method = 1 : i32, id = "keeping_id", max_percentile = 0.000000e+00 : f32, min_percentile = 0.000000e+00 : f32, num_bins = 0 : i32}> : (tensor<10x1x3xf32>) -> (tensor<10x1x3xf32>, tensor<f32>, tensor<f32>, tensor<0xi64>)
    return %output_0 : tensor<10x1x3xf32>
  }
  // CHECK-LABEL: @main

// CHECK:           [[VAL_12:%.*]]:4 = "tf.SplitV"([[VAL_7]], [[VAL_10]], [[VAL_11]]) : (tensor<40x8xf32>, tensor<4xi32>, tensor<i32>) -> (tensor<10x8xf32>, tensor<10x8xf32>, tensor<10x8xf32>, tensor<10x8xf32>)
// CHECK-DAG:       [[VAL_13:%.*]] = "tf.Const"() <{value = dense<10> : tensor<4xi32>}> : () -> tensor<4xi32>

    %output_0, %min_1, %max_2, %histogram_3 = "tf.CustomAggregator"(%0) <{calibration_method = 1 : i32, id = "1", max_percentile = 0.000000e+00 : f32, min_percentile = 0.000000e+00 : f32, num_bins = 0 : i32}> : (tensor<10x1x3xf32>) -> (tensor<10x1x3xf32>, tensor<f32>, tensor<f32>, tensor<0xi64>)
    return %output_0 : tensor<10x1x3xf32>
  }
  // CHECK-LABEL: @main

    %1 = stablehlo.maximum %0, %cst : tensor<10x1x3xf32>
    return %1 : tensor<10x1x3xf32>
  }
}

// -----

module attributes {tf.versions = {bad_consumers = [], min_consumer = 12 : i32, producer = 1836 : i32}, tf_saved_model.semantics} {

// CHECK: return %[[RES]]
}

func.func @scatterNdHigherRankIndices(%arg0: tensor<4x2x2xi32>, %arg1: tensor<4x2x3xf32>, %arg2: tensor<3xi32>) -> tensor<10x2x3xf32> {
  %0 = "tf.ScatterNd"(%arg0, %arg1, %arg2) : (tensor<4x2x2xi32>, tensor<4x2x3xf32>, tensor<3xi32>) -> tensor<10x2x3xf32>
  func.return %0 : tensor<10x2x3xf32>

// CHECK-LABEL:scatterNdHigherRankIndices

  %9 = "mhlo.compare"(%0, %8) {comparison_direction = #mhlo<comparison_direction NE>} : (tensor<10x8xf32>, tensor<10x8xf32>) -> tensor<10x8xi1>
  %10 = mhlo.and %7, %9 : tensor<10x8xi1>
  %11 = mhlo.subtract %arg0, %6 : tensor<10x8xf32>
  %12 = mhlo.divide %11, %5 : tensor<10x8xf32>
  %13 = mhlo.add %12, %3 : tensor<10x8xf32>

  %push = "tf.XlaDynamicUpdateSlice"(%tl, %arg0_reshaped, %start_indices) : (tensor<10x8x4xf32>, tensor<1x8x4xf32>, tensor<3xi32>) -> tensor<10x8x4xf32>
  %one = "tf.Const"() {value = dense<1> : tensor<1xi32>} : () -> tensor<1xi32>
  %next_index_in_list = "tf.AddV2"(%index_in_list, %one) : (tensor<1xi32>, tensor<1xi32>) -> tensor<1xi32>
  return

      // were:
      //   tensor<!tf_type.variant<tensor<?x8xf32>>>
      // and:
      //   tensor<!tf_type.variant<tensor<10x8xf32>>>
      // we'll try here to refine tensor<?x8xf32> with tensor<10x8xf32>.
      auto refined_subtype = mlir::cast<TensorType>(
          TypeMeet(lhs_element_type_with_subtype.GetSubtypes().front(),

// CHECK:  return %[[RESULT]]
}

func.func @ConvertSqueezeToReshapeWithDynamicDimension2(%arg0: tensor<?x1x8x3xf32>) -> tensor<1x8x3xf32> {
  %0 = "tfl.squeeze"(%arg0) {squeeze_dims = [0]}: (tensor<?x1x8x3xf32>) -> tensor<1x8x3xf32>
  func.return %0: tensor<1x8x3xf32>

// CHECK-LABEL: ConvertSqueezeToReshapeWithDynamicDimension2
// CHECK-DAG: [[CONST:.*]] = arith.constant dense<[1, 8, 3]> : tensor<3xi32>