%7 = call @uniform_quantize_0(%arg0, %0, %1) : (tensor<1x2xf32>, tensor<1x1xf32>, tensor<1x1xi8>) -> tensor<1x2xi8>
    %8 = stablehlo.convert %7 : (tensor<1x2xi8>) -> tensor<1x2xf32>
    %9 = stablehlo.convert %2 : (tensor<2x3xi8>) -> tensor<2x3xf32>
    %10 = stablehlo.dot_general %8, %9, contracting_dims = [1] x [0] : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>

  %2:2 = "tf.IteratorGetNext"(%arg0) {device = "/device:CPU:0"} : (tensor<*x!tf_type.resource>) -> (tensor<128xf32>, tensor<128xf32>)
  %3:2 = "tf.IteratorGetNext"(%arg1) {device = "/device:CPU:0"} : (tensor<*x!tf_type.resource>) -> (tensor<128xf32>, tensor<128xf32>)
  // CHECK: "tf.TPUCompileSucceededAssert"(%[[COMPILE]]#0)
  "tf_device.launch"() ({

// CHECK-DAG: %[[input_19:.*]] = "quantfork.stats"({{.*}}) <{layerStats = dense<[-2.000000e+00, 4.000000e+00]> : tensor<2xf32>}> : (tensor<1x2xf32>) -> tensor<1x2xf32>

// CHECK: "tfl.unidirectional_sequence_lstm"({{.*}}, %[[input_18]], %[[input_19]], %{{[0-9]+}}, %{{[0-9]+}}, %{{[0-9]+}}, %{{[0-9]+}})

}

// CHECK-LABEL: QuantizeConcat
func.func @QuantizeConcat(tensor<1x2xf32>, tensor<1x2xf32>) -> tensor<2x2x!quant.uniform<u8:f32, 1.000000e-01:128>> {
^bb0(%arg0: tensor<1x2xf32>, %arg1: tensor<1x2xf32>):
  %0 = "tfl.concatenation"(%arg0, %arg1) {axis = 0 : i32, fused_activation_function = "NONE"} : (tensor<1x2xf32>, tensor<1x2xf32>) -> tensor<2x2xf32>

  %0 = "tf_device.cluster_func"(%arg0, %arg1) {func = @_func, use_spmd_for_xla_partitioning = true, use_tpu = true, num_cores_per_replica = 2 : i64} : (tensor<2x4xf32>, tensor<4x2xf32>) -> tensor<2x2xf32>
  %1:2 = "tf.TPUPartitionedOutputV2"(%0) {device = "", partition_dims = [2, 1]} : (tensor<2x2xf32>) -> (tensor<1x2xf32>, tensor<1x2xf32>)

}

// -----

func.func @testConcatV2(%arg0: tensor<8x16xf32>, %arg1: tensor<16x8xf32>, %axis: tensor<i32>) -> tensor<*xf32> {
  // expected-error @+1 {{operand type 'tensor<16x8xf32>' is not compatible with preceding operands; expected dimension at index 1: 16}}
  %0 = "tf.ConcatV2"(%arg0, %arg1, %axis) : (tensor<8x16xf32>, tensor<16x8xf32>, tensor<i32>) -> tensor<*xf32>
  func.return %0 : tensor<*xf32>
}

// -----

  // CHECK-DAG: %[[ITEMS0_0:.*]] = "tf.ExpandDims"(%[[ITEMS0]], %[[AXIS]])
  // CHECK-DAG: "tf.ConcatV2"(%[[ITEMS1_3]], %[[ITEMS1_2]], %[[ITEMS1_1]], %[[ITEMS1_0]], %[[ITEMS0_0]], %[[AXIS]]) : (tensor<1x2xf32>, tensor<1x2xf32>, tensor<1x2xf32>, tensor<1x2xf32>, tensor<1x2xf32>, tensor<i64>) -> tensor<5x2xf32>

  %indices0 = "tf.Const"() {value = dense<4> : tensor<i32>} : () -> tensor<i32>

  %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>

// CHECK-LABEL: testFoldEnsureShapeOp
func.func @testFoldEnsureShapeOp(%arg0: tensor<10x20xf32>) -> (tensor<10x20xf32>, tensor<10x20xf32>, tensor<20x10xf32>) {
  %0 = "tf.EnsureShape"(%arg0) {shape = #tf_type.shape<10x20>} : (tensor<10x20xf32>) -> tensor<10x20xf32>
  %1 = "tf.EnsureShape"(%arg0) {shape = #tf_type.shape<?x20>} : (tensor<10x20xf32>) -> tensor<10x20xf32>
  // Failing case which should not be folded.

)mlir";

constexpr absl::string_view kModuleHybridQuantized = R"mlir(
  module {
    func.func @main(%arg0: tensor<1x2xf32>, %arg1: tensor<2x3x!quant.uniform<i8:f32, 6.000000e-03:0>> {tf_saved_model.index_path = ["input_tensor"]}) -> (tensor<1x3xf32>) {
      %0 = stablehlo.dot_general %arg0, %arg1, contracting_dims = [1] x [0] : (tensor<1x2xf32>, tensor<2x3x!quant.uniform<i8:f32, 6.000000e-03:0>>) -> tensor<1x3xf32>
      return %0 : tensor<1x3xf32>
    }