// CHECK-NEXT:     %[[GROUP:.*]] = "tf.Const"()
  // CHECK-SAME:       [0, 1, 2, 3]
  // CHECK-NEXT:     %[[REDUCED:.*]] = "tf.XlaAllReduce"(%[[ID]], %[[GROUP]]) <{mode = "CrossReplica", reduce_op = "Add"}> : (tensor<f32>, tensor<1x4xi32>) -> tensor<f32>
  // CHECK-NEXT:     "tf.OpA"(%[[REDUCED]]) : (tensor<f32>) -> ()
  tf_device.replicate {n = 4 : i32} {
    "tf_device.cluster"() ({
      "tf.OpA"(%arg0) : (tensor<f32>) -> ()
      tf_device.return

    value {
      type: DT_BOOL
    }
  }
}

# CHECK-DAG: tf.IteratorGetNext{{.+}}-> (tensor<1x8xbf16>, tensor<2x?x16xf32>, tensor<*xf64>)
# CHECK-DAG: tf.IteratorGetNextSync{{.+}}-> (tensor<*xi16>, tensor<3x24xi32>, tensor<?x4x32xi64>)
# CHECK-DAG: tf.MultiDeviceIteratorGetNextFromShard{{.+}}-> (tensor<5x40xf16>, tensor<*xcomplex<f32>>, tensor<6x48x?xcomplex<f64>>)

// Note: Argument quantization sequence omitted.
// CHECK: stablehlo.dot_general %{{.+}}, %{{.+}}, contracting_dims = [1] x [0] : (tensor<1x1024xi8>, tensor<1024x3xi8>) -> tensor<1x3xi32>

// Note: Result dequantization sequence omitted.
// CHECK: return %{{.+}} : tensor<1x3xf32>

// -----

// Tests that a simple dot_general (lifted as a function) with CustomAggregators