%2 = "quantfork.stats"(%1) {layerStats = dense<[5.00000000e-6, 7.00000000e-1]> : tensor<2xf32>} : (tensor<2x2x3xf32>) -> tensor<2x2x3xf32>
    return %2 : tensor<2x2x3xf32>
  }
// CHECK: func.func private @quantize_dot_general_batch_per_tensor_quantized_fn(%[[ARG_0:.+]]: tensor<2x2x2xf32>) -> tensor<2x2x3xf32> attributes {tf._original_func_name = "main_0"}

  %cst_1 = arith.constant dense<6.000000e+00> : tensor<2x1x3x3xf32>
  %0 = "tfl.quantize"(%cst_1) {qtype = tensor<2x1x3x3x!quant.uniform<i8<-127:127>:f32:0, {6.587140e-03,1.888450e-02}>>} : (tensor<2x1x3x3xf32>) -> tensor<2x1x3x3x!quant.uniform<i8<-127:127>:f32:0, {6.587140e-03,1.888450e-02}>>
  %1 = "tfl.dequantize"(%0) : (tensor<2x1x3x3x!quant.uniform<i8<-127:127>:f32:0, {6.587140e-03,1.888450e-02}>>) -> tensor<2x1x3x3xf32>

    %10 = "tf.AddV2"(%9, %cst_2) {device = ""} : (tensor<2x1x3xi32>, tensor<3xi32>) -> tensor<2x1x3xi32>
    %11 = "tf.Cast"(%10) {Truncate = false, device = ""} : (tensor<2x1x3xi32>) -> tensor<2x1x3xf32>
    %12 = "tf.Mul"(%11, %cst) {device = ""} : (tensor<2x1x3xf32>, tensor<f32>) -> tensor<2x1x3xf32>
    %13 = "tf.Identity"(%12) {device = ""} : (tensor<2x1x3xf32>) -> tensor<2x1x3xf32>

func.func @QuantizePad2(tensor<2x1x3x!quant.uniform<u8:f32, 0.1>>, tensor<2x1x3xf32>, tensor<3x2xi32>) -> (tensor<?xf32>, tensor<?xf32>) {
^bb0(%arg0: tensor<2x1x3x!quant.uniform<u8:f32, 0.1>>, %arg1: tensor<2x1x3xf32>, %arg2: tensor<3x2xi32>):
  %0 = "tfl.dequantize"(%arg0) : (tensor<2x1x3x!quant.uniform<u8:f32, 0.1>>) -> tensor<2x1x3xf32>
  %1 = "tfl.pad"(%arg1, %arg2) : (tensor<2x1x3xf32>, tensor<3x2xi32>) -> tensor<?xf32>

}

// CHECK-LABEL: testConvertPackToReshapeAxis1
func.func @testConvertPackToReshapeAxis1(%arg0: tensor<2x3xf32>) -> tensor<2x1x3xf32> {
  %0 = "tf.Pack"(%arg0) {axis = 1 : i64, device = "/job:localhost/replica:0/task:0/device:GPU:0"} : (tensor<2x3xf32>) -> tensor<2x1x3xf32>
  func.return %0 : tensor<2x1x3xf32>
  // CHECK: %[[SHAPE:.*]] = "tf.Const"() <{value = dense<[2, 1, 3]> : tensor<3xi32>}> : () -> tensor<3xi32>

  %0 = "mhlo.broadcast_in_dim"(%arg0) <{broadcast_dimensions = dense<[1, 2, 3]> : tensor<3xi64>, name = "broadcast.0"}> : (tensor<8x1x16xf32>) -> tensor<3x8x8x16xf32>
  func.return %0 : tensor<3x8x8x16xf32>
}

// CHECK-LABEL:   func @broadcast_in_dim_general_case(
// CHECK-SAME:                                        %[[VAL_0:.*]]: tensor<3x1x16xf32>) -> tensor<3x8x8x16xf32> {

// -----

func.func @testBatchMatMulV2InvalidBroadcastingBatchDimensionWithHigherYRank(%lhs: tensor<2x5x10xf32>, %rhs: tensor<10x10x10x10xf32>) {
  // expected-error @+1 {{found incompatible broadcast batch dimensions for lhs shape 'tensor<2x5x10xf32>' and rhs shape 'tensor<10x10x10x10xf32>'}}
  %0 = "tf.BatchMatMulV2"(%lhs, %rhs) : (tensor<2x5x10xf32>, tensor<10x10x10x10xf32>) -> tensor<10x10xf32>
}

// -----

  func.return %0 : tensor<2x2xf16>
}

// -----

// CHECK-LABEL: squeeze
func.func @squeeze(%arg0: tensor<1x1x10xf32>) -> tensor<1x10xf32> {
  // CHECK: mhlo.reshape
  %0 = "tf.Squeeze"(%arg0) : (tensor<1x1x10xf32>) -> tensor<1x10xf32>
  func.return %0 : tensor<1x10xf32>
}

// -----

// CHECK-LABEL: squeeze_ranked