%t2 = "tf.Acosh"(%arg1) : (tensor<2xf32>) -> tensor<2xf32>
    "tf.Yield"(%t0, %t1, %t2) : (tensor<2xf32>, tensor<2xf32>, tensor<2xf32>) -> ()
    }, {
     %e0 = "tf.Neg"(%arg1) : (tensor<2xf32>) -> tensor<2xf32>
     %e1 = "tf.Relu"(%arg1) : (tensor<2xf32>) -> tensor<2xf32>
     %e2 = "tf.Sin"(%arg1) : (tensor<2xf32>) -> tensor<2xf32>
     "tf.Yield"(%e0, %e1, %e2) : (tensor<2xf32>, tensor<2xf32>, tensor<2xf32>) -> ()

    version = 5 : i64
  } : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>
  return %0 : tensor<1x3xf32>
}

// CHECK-LABEL: func.func @qdq_for_dot_general_weight_empty
// CHECK-SAME: (%[[ARG_0:.+]]: tensor<1x2xf32>) -> tensor<1x3xf32>
// CHECK: %[[CST:.+]] = "tf.Const"() <{value = dense<3.000000e-01> : tensor<2x3xf32>}> : () -> tensor<2x3xf32>

    %9 = stablehlo.convert %2 : (tensor<2x3xi32>) -> tensor<2x3xf32>
    %10 = stablehlo.dot_general %8, %9, contracting_dims = [1] x [0] : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>
    %11 = stablehlo.convert %3 : (tensor<1x3xi32>) -> tensor<1x3xf32>
    %12 = stablehlo.subtract %10, %11 : tensor<1x3xf32>  // q1 * q2 - z1 * q2
    %13 = stablehlo.multiply %12, %4 : tensor<1x3xf32>  // s1 * s2

    %17 = "tf.Mul"(%16, %cst_0) : (tensor<2x2xf32>, tensor<f32>) -> tensor<2x2xf32>
    %18 = "tf.AddV2"(%17, %cst) : (tensor<2x2xf32>, tensor<f32>) -> tensor<2x2xf32>
    %19 = "tf.Floor"(%18) : (tensor<2x2xf32>) -> tensor<2x2xf32>
    %20 = "tf.ClipByValue"(%19, %cst_5, %cst_6) : (tensor<2x2xf32>, tensor<f32>, tensor<f32>) -> tensor<2x2xf32>

// CHECK-LABEL: hardcode_all
func.func @hardcode_all(%arg0: tensor<2x2xf32>, %arg1: tensor<2x1xf32>) -> tensor<2x2xf32> {
  %0 = "tfl.add"(%arg0, %arg1) {fused_activation_function="NONE"}: (tensor<2x2xf32>, tensor<2x1xf32>) -> tensor<2x2xf32>
  func.return %0 : tensor<2x2xf32>

// CHECK: %[[q0:.*]] = "tfl.quantize"(%arg1) <{qtype = tensor<2x1x!quant.uniform<u8:f32, 0.0078431372549019607:128>>}>

}
func.func @_func(%arg0: tensor<2x4xf32>, %arg1: tensor<4x2xf32>) -> tensor<2x2xf32> {
  %0 = "tf.MatMul"(%arg0, %arg1) {_XlaSharding = "\08\03\1A\02\02\01\22\02\00\01"} : (tensor<2x4xf32>, tensor<4x2xf32>) -> tensor<2x2xf32>
  %1 = "tf.Identity"(%0) : (tensor<2x2xf32>) -> tensor<2x2xf32>
  return %1 : tensor<2x2xf32>
}

// -----
// The following op sharding is used in the following test case:

  // CHECK: return %[[MATMUL_1]] : tensor<4x6xf32>
}

// -----

func.func @batchMatMulMatrixAdjXY(%arg0: tensor<5x4xf32>, %arg1: tensor<6x5xf32>) -> tensor<4x6xf32> {
  %0 = "tf.BatchMatMul"(%arg0, %arg1) {adj_x = true, adj_y = true} : (tensor<5x4xf32>, tensor<6x5xf32>) -> tensor<4x6xf32>
  func.return %0 : tensor<4x6xf32>

    return %1 : tensor<1x3xf32>
  }

  func.func private @composite_dot_general_fn(%arg0: tensor<1x2xf32>, %arg1: tensor<2x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module} {
    %0 = stablehlo.dot_general %arg0, %arg1, contracting_dims = [1] x [0] : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>
    return %0 : tensor<1x3xf32>
  }
}

func.func @testSubOfZero(%arg0: tensor<?x1xf32>, %arg1: tensor<4x1xf32>) -> (tensor<?x1xf32>, tensor<4x1xf32>) {
  %0 = "tf.Const"() {value = dense<0.0> : tensor<f32>} : () -> tensor<f32>
  %1 = "tf.Sub"(%arg0, %0) : (tensor<?x1xf32>, tensor<f32>) -> tensor<?x1xf32>
  %2 = "tf.Sub"(%arg1, %0) : (tensor<4x1xf32>, tensor<f32>) -> tensor<4x1xf32>
  func.return %1, %2: tensor<?x1xf32>, tensor<4x1xf32>

// CHECK: return %arg0, %arg1
}

}

// CHECK-LABEL: prepareAdd
func.func @prepareAdd(%arg0: tensor<2x2xf32>) -> tensor<2x2xf32> {
  %cst = arith.constant dense<[[0.0, 1.0], [2.0, 255.0]]> : tensor<2x2xf32>
  %add = "tfl.add"(%arg0, %cst) {fused_activation_function="NONE"} : (tensor<2x2xf32>, tensor<2x2xf32>) -> tensor<2x2xf32>
  func.return %add : tensor<2x2xf32>