// CHECK: %[[RES:.*]] = chlo.broadcast_add %[[LHS_2]], %[[RHS_2]] {broadcast_dimensions = array<i64: 1>} :
  // CHECK-SAME: (tensor<3x2x!quant.uniform<i32:f32, 2.000000e+00:4>>, tensor<2x!quant.uniform<i32:f32, 2.000000e+00:4>>)
  // CHECK-SAME: -> tensor<3x2x!quant.uniform<i32:f32, 2.000000e+00:4>>
  %result = "tf.UniformQuantizedAdd"(
    %input1_qint, %input2_qint,
    %input_scales, %input_zps,
    %bias_scales, %bias_zps,

  // CHECK-DAG: %[[CONVERT_1:.*]] = mhlo.bitcast_convert %[[LHS]] : (tensor<?x!quant.uniform<i8:f32, 2.000000e+00:4>>) -> tensor<?xi8>
  // CHECK-DAG: %[[CONVERT_2:.*]] = mhlo.bitcast_convert %[[CONVERT_1]] : (tensor<?xi8>) -> tensor<?x!quant.uniform<i8:f32, 2.000000e+00:4>>

func.func @foldBroadcastInDimBeforeMulOp_bcast_dim_2D_float() -> (tensor<1x2x2x3xf32>) {
  // CHECK-DAG: %[[RES:.*]] = mhlo.constant dense<{{\[\[\[\[}}1.000000e+00, 4.000000e+00, 9.000000e+00], [4.000000e+00, 1.000000e+01, 1.800000e+01]], {{\[\[}}2.800000e+01, 4.000000e+01, 5.400000e+01], [4.000000e+01, 5.500000e+01, 7.200000e+01]]]]> : tensor<1x2x2x3xf32>

  func.return %1 : tensor<256x8x7x16xf16>

  // CHECK-DAG: %cst = arith.constant dense<[2.500000e+00, 3.500000e+00, 4.500000e+00, 5.500000e+00, 6.500000e+00, 7.500000e+00, 8.500000e+00, 9.500000e+00, 1.050000e+01, 1.150000e+01, 1.250000e+01, 1.350000e+01, 1.450000e+01, 1.550000e+01, 1.650000e+01, 1.750000e+01]> : tensor<16xf16>

// 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-SAME: %[[INPUT:[^:[:space:]]+]]
func.func @fuseMulAndConv2D(%input: tensor<1x256x256x3xf32>) -> (tensor<1x256x256x2xf32>) {
  // CHECK-DAG: %[[FILTER:.+]] = mhlo.constant dense<{{\[\[\[\[}}1.000000e+00, 2.000000e+00], [3.000000e+00, 4.000000e+00], [5.000000e+00, 6.000000e+00]]]]> : tensor<1x1x3x2xf32>
  // CHECK-DAG: %[[CST:.+]] = mhlo.constant dense<[1.000000e-01, 2.000000e-01]> : tensor<2xf32>

  // CHECK: value = dense<[1.000000e+00, 2.000000e+00, 3.000000e+00, 4.000000e+00]> : tensor<4xf16>
  %0 = "tfl.pseudo_const"() { value = dense<[1.0, 2.0, 3.0, 4.0]> : tensor<4xf16> } : () -> tensor<4xf16>
  func.return %0 : tensor<4xf16>
}

func.func @f32() -> tensor<4xf32> {
  // CHECK-LABEL: @f32
  // CHECK: value = dense<[1.000000e+00, 2.000000e+00, 3.000000e+00, 4.000000e+00]> : tensor<4xf32>

  // CHECK: value = dense<[1.000000e+00, 2.000000e+00, 3.000000e+00, 4.000000e+00]> : tensor<4xf16>
  %0 = "tfl.pseudo_const"() { value = dense<[1.0, 2.0, 3.0, 4.0]> : tensor<4xf16> } : () -> tensor<4xf16>
  func.return %0 : tensor<4xf16>
}

func.func @f32() -> tensor<4xf32> {
  // CHECK-LABEL: @f32
  // CHECK: value = dense<[1.000000e+00, 2.000000e+00, 3.000000e+00, 4.000000e+00]> : tensor<4xf32>

  %1 = stablehlo.transpose %0, dims = [1, 0] : (tensor<2x3xf32>) -> tensor<3x2xf32>
  return %1 : tensor<3x2xf32>
}
// CHECK-DAG: %[[CONST_0:.+]] = stablehlo.constant dense<{{\[\[}}0.000000e+00, 3.000000e+00], [1.000000e+00, 4.000000e+00], [2.000000e+00, 5.000000e+00]]> : tensor<3x2xf32>
// CHECK-NOT: transpose
// CHECK: return %[[CONST_0]] : tensor<3x2xf32>

// -----

// CHECK-LABEL: transpose_simple_4d

  // CHECK-DAG: %[[LOOKUP1:.*]] = arith.constant dense<{{\[\[\[}}1.000000e+00, 2.000000e+00], [3.000000e+00, 4.000000e+00]], {{\[\[}}5.000000e+00, 6.000000e+00], [7.000000e+00, 8.000000e+00]], {{\[\[}}1.000000e+00, 2.000000e+00], [3.000000e+00, 4.000000e+00]]]> : tensor<3x2x2xf32>