break;
      case COSH:
        y = Cosh(scope_, x);
        break;
      case TANH:
        y = Tanh(scope_, x);
        break;
      case ASINH:
        y = Asinh(scope_, x);
        break;
      case ACOSH:
        y = Acosh(scope_, x);
        break;
      case ATANH:
        y = Atanh(scope_, x);
        break;
      case SIGMOID:
        y = Sigmoid(scope_, x);
        break;

}

func TestAtanh(t *testing.T) {
	for i := 0; i < len(vf); i++ {
		a := vf[i] / 10
		if f := Atanh(a); !veryclose(atanh[i], f) {
			t.Errorf("Atanh(%g) = %g, want %g", a, f, atanh[i])
		}
	}
	for i := 0; i < len(vfatanhSC); i++ {
		if f := Atanh(vfatanhSC[i]); !alike(atanhSC[i], f) {
			t.Errorf("Atanh(%g) = %g, want %g", vfatanhSC[i], f, atanhSC[i])
		}
	}
}

func TestAtan2(t *testing.T) {

  // CHECK: %[[ARG_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%arg0, %[[ARG_PERM]])
  // CHECK: %[[TANH:[0-9]*]] = "tf.Tanh"(%[[ARG_TRANSPOSE]]) {{.*}} tensor<1x8x4x4xf32>
  // CHECK: %[[ADD:[0-9]*]] = "tf.AddV2"(%[[TANH]], %[[TANH]]) {{.*}} tensor<1x8x4x4xf32>
  // CHECK: return %[[ADD]]

  %0 = "tf.Tanh"(%arg0) : (tensor<1x4x4x8xf32>) -> tensor<1x4x4x8xf32>

  elif act == 'TANH':
    return tf.raw_ops.Tanh(x=res)
  else:
    return res


@tf.RegisterGradient('NewConv2D')
def _conv_add_relu_grad(op: ops.Operation, grad):
  act = op.get_attr('act')
  y = op.outputs[0]
  if act == 'RELU':
    grad = gen_nn_ops.relu_grad(grad, y)
  elif act == 'RELU6':
    grad = gen_nn_ops.relu6_grad(grad, y)
  elif act == 'TANH':
    y = math_ops.conj(y)

  // CHECK: %[[RES_PERM:.*]] = "tf.Const"() <{value = dense<[0, 3, 1, 2]> : tensor<4xi32>}>
  // CHECK: %[[TANH:[0-9]*]] = "tf.Tanh"(%arg0) {{.*}} tensor<1x4x4x8xf32>
  // CHECK: %[[RES_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%[[TANH]], %[[RES_PERM]]) {{.*}} tensor<1x8x4x4xf32>
  // CHECK: return %[[RES_TRANSPOSE]]

  %0 = "tf.Const"() {value = dense<[0, 3, 1, 2]> : tensor<4xi32>} : () -> tensor<4xi32>

    %5, %6, %7, %8,
    %9, %9, %9,
    %10, %11,
    %10, %10,
    %9, %9,
    %recurrent_input, %cell_input,
    %9, %9, %9, %9) {
      cell_clip = 1.000000e+01 : f32,
      fused_activation_function = "TANH",
      proj_clip = 0.000000e+00 : f32,
      time_major = false} : (
        tensor<1x2x3xf32>,
        tensor<1x1xf32>, tensor<1x1xf32>, tensor<1x1xf32>, tensor<1x1xf32>,

  // CHECK-NEXT: [[r2:%.*]] = tfrt_fallback_async.executeop {{.*}} "tf.Tanh"([[r1]]) {T = f32}
  %4 = "tf.Tanh"(%3) {T = f32, _output_shapes = ["tfshape$dim { size: 3 } dim { size: 3 }"], device = "/device:CPU:0"} : (tensor<3x3xf32>) -> tensor<3x3xf32>
  // CHECK-NOT: tf.Identity

  ops::Switch switch_b(root.WithOpName("switch_b"), value, cond_b);

  Output tanh_a0 = ops::Tanh(root.WithOpName("tan_a0"), switch_a.output_true);
  Output tanh_a1 = ops::Tanh(root.WithOpName("tan_a1"), tanh_a0);

  Output tanh_b0 = ops::Tanh(root.WithOpName("tan_b0"), switch_b.output_true);
  Output tanh_b1 = ops::Tanh(root.WithOpName("tan_b1"), tanh_b0);

  Output add = ops::Add(root.WithOpName("add"), tanh_a1, tanh_b1);

    %9, %9, %9, %9) {
      asymmetric_quantize_inputs = false,
      cell_clip = 1.000000e+01 : f32,
      effective_hidden_scale_intermediate = tensor<0x!quant.calibrated<f32<0.0:1.0>>>,
      fused_activation_function = "TANH",
      input_to_cell_intermediate = tensor<0xf32>,
      input_to_forget_intermediate = tensor<0xf32>,
      input_to_input_intermediate = tensor<0xf32>,
      input_to_output_intermediate = tensor<0xf32>,

"tfl.lstm"(%arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %cst, %cst, %cst, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg19, %arg20, %arg15, %arg16, %arg17, %arg18) ({}) {cell_clip = 1.000000e+01 : f32, fused_activation_function = "TANH", input_to_input_intermediate = tensor<0x!quant.uniform<i16:f32, 0.0049890000373125076>>, input_to_forget_intermediate = tensor<0x!quant.uniform<i16:f32, 0.0078849997371435165>>, input_to_cell_intermediate = tensor<0x!quant.uniform<i16:f32, 0.0087630003690719604>>,...