auto matmul_result =
        BatchMatMul(scope, grad_grad_expand, logits_softmax_expand);
    axis.push_back(1);
    auto squeeze_result = Squeeze(scope, matmul_result, Squeeze::Axis(axis));
    auto subtraction_result = Subtract(scope, grad_grad, squeeze_result);
    auto multiply_result = Multiply(scope, subtraction_result, logits_softmax);
    grad = Add(scope, grad, multiply_result);

 Fill 1
 GreaterEqual 8
 Identity 1
 Less 1
 MatMul 10
 Mul 44
 Range 1
 Rsqrt 1
 Select 19
 Shape 6
 Sigmoid 24
 Snapshot 8
 Softmax 1
 Split 8
 SplitV 6
 Square 1
 Squeeze 1
 StridedSlice 1
 Sum 2
 Tanh 17
cluster 12 size 6
 Add 1
 All 1
 Const 2
 GreaterEqual 1
 LogicalOr 1
cluster 15 size 614
 Add 22
 AddN 41
 BatchMatMulV2 2
 BiasAddGrad 8

			hostTests: []nssHostTest{
				{"x.com", "myhostname", hostLookupFilesDNS},
				{"somehostname", "myhostname", hostLookupFilesDNS},
				{"myhostname", "myhostname", hostLookupCgo},
			},
		},
		// Debian Squeeze is just "dns,files", but lists all
		// the default criteria for dns, but then has a
		// non-standard but redundant notfound=return for the
		// files.
		{
			name:   "debian_squeeze",
			c:      &conf{},

        expanded_fft_len.getResult());

    // Insert the squeeze op.
    auto squeeze_dim = rewriter.getI64ArrayAttr({-2});
    TF::SqueezeOp squeeze = rewriter.create<TF::SqueezeOp>(
        rfft_op.getLoc(), output_type, rfft2d.getResult(), squeeze_dim);

    rewriter.replaceOp(op, squeeze.getResult());

    return success();
  }
};

  %0 = "tf.Squeeze"(%arg0) : (tensor<1x2x2xf32>) -> tensor<2x2xf32>
  func.return %0 : tensor<2x2xf32>

// CHECK-LABEL:squeezeDefault
// CHECK:  "tfl.squeeze"(%arg0) <{squeeze_dims = []}> : (tensor<1x2x2xf32>) -> tensor<2x2xf32>
}

func.func @squeezeSingleAxis(%arg0: tensor<2x1x2xf32>) -> tensor<2x2xf32> {
  %0 = "tf.Squeeze"(%arg0) {squeeze_dims = [1]} : (tensor<2x1x2xf32>) -> tensor<2x2xf32>

  %outputs_20 = "tf.ResizeBilinear"(%outputs_18, %outputs) {align_corners = false, device = "", half_pixel_centers = false} : (tensor<1x?x?x3xui8>, tensor<2xi32>) -> tensor<1x224x224x3xf32>
  %outputs_22 = "tf.Squeeze"(%outputs_20) {device = "", squeeze_dims = [0]} : (tensor<1x224x224x3xf32>) -> tensor<224x224x3xf32>
  %outputs_24 = "tf.Cast"(%outputs_22) {Truncate = false, device = ""} : (tensor<224x224x3xf32>) -> tensor<224x224x3xui8>

}

TEST_F(ArrayGradTest, SqueezeGrad) {
  TensorShape x_shape({1, 5, 1, 2});
  auto x = Placeholder(scope_, DT_FLOAT, Placeholder::Shape(x_shape));
  TensorShape y_shape({5, 2});
  auto y = Squeeze(scope_, x);
  RunTest(x, x_shape, y, y_shape);
}

TEST_F(ArrayGradTest, TransposeGrad) {
  TensorShape x_shape({5, 2});
  auto x = Placeholder(scope_, DT_FLOAT, Placeholder::Shape(x_shape));

    %12 = "tf.Cast"(%arg1) {Truncate = false} : (tensor<2x1xf32>) -> tensor<2x1xi64>
    %13 = "tf.Reshape"(%12, %9) : (tensor<2x1xi64>, tensor<1xi32>) -> tensor<2xi64>
    %14 = "tf.Squeeze"(%arg1) {squeeze_dims = [-1]} : (tensor<2x1xf32>) -> tensor<2xf32>
    // CHECK: "tf.Conv2D"
    // CHECK-SAME: strides = [1, 1, 1, 1]

  %267 = "tf.Reshape"(%266, %5) {device = ""} : (tensor<?x?xi1>, tensor<1xi32>) -> tensor<?xi1>
  %268 = "tf.Where"(%267) {device = ""} : (tensor<?xi1>) -> tensor<?x1xi64>
  %269 = "tf.Squeeze"(%268) {device = "", squeeze_dims = [1]} : (tensor<?x1xi64>) -> tensor<?xi64>
  %270 = "tf.GatherV2"(%264, %269, %14) {batch_dims = 0 : i64, device = ""} : (tensor<?xi64>, tensor<?xi64>, tensor<i32>) -> tensor<?xi64>

  PAD = 34,
  UNIDIRECTIONAL_SEQUENCE_RNN = 35,
  GATHER = 36,
  BATCH_TO_SPACE_ND = 37,
  SPACE_TO_BATCH_ND = 38,
  TRANSPOSE = 39,
  MEAN = 40,
  SUB = 41,
  DIV = 42,
  SQUEEZE = 43,
  UNIDIRECTIONAL_SEQUENCE_LSTM = 44,
  STRIDED_SLICE = 45,
  BIDIRECTIONAL_SEQUENCE_RNN = 46,
  EXP = 47,
  TOPK_V2 = 48,
  SPLIT = 49,
  LOG_SOFTMAX = 50,