func.return %3 : tensor<10x10x10xf32>
  // CHECK:  return %arg0
}

func.func @NotFoldStridedSlice(%arg0 : tensor<10x10x10xf32>) -> tensor<9x9x9xf32> {
  %0 = "tfl.pseudo_const"() {value = dense<[9, 9, 9]> : tensor<3xi32>} : () -> tensor<3xi32>

      rhs_contracting_dimensions = [0]
    >,
    precision_config = [#mhlo<precision DEFAULT>, #mhlo<precision DEFAULT>]
  } : (tensor<1x1x1024xf32>, tensor<1024x1024xf32>) -> tensor<1x1x1024xf32>
  func.return %0 : tensor<1x1x1024xf32>

// CHECK-LABEL:   convert_dot_general_repeated
// CHECK:         %[[RESHAPED_0:.*]] = mhlo.reshape %arg0
// CHECK-NEXT:    %[[RESHAPED_1:.*]] = mhlo.reshape %arg1

// CHECK: %[[v0:.*]] = "tf.Reshape"(%arg0, %[[cst]]) : (tensor<2x1x1x11xf32>, tensor<3xi64>) -> tensor<2x1x11xf32>
// CHECK: %[[v1:.*]] = "tf.BatchMatMulV2"(%[[v0]], %arg1) <{adj_x = false, adj_y = false}> : (tensor<2x1x11xf32>, tensor<2x11x2xf32>) -> tensor<2x1x2xf32>
// CHECK: %[[v2:.*]] = "tf.Reshape"(%[[v1]], %[[cst_1]]) : (tensor<2x1x2xf32>, tensor<4xi64>) -> tensor<2x1x1x2xf32>

  // expected-error @+1 {{found incompatible broadcast batch dimensions for lhs shape 'tensor<10x2x5x10xf32>' and rhs shape 'tensor<10x10x10xf32>'}}
  %0 = "tf.BatchMatMulV2"(%lhs, %rhs) : (tensor<10x2x5x10xf32>, tensor<10x10x10xf32>) -> tensor<10x10xf32>
}

// -----

  return %2 : tensor<1x1x1x4xf32>
}
func.func private @XlaCallModule_aten.avg_pool2d.default.impl_2(%arg0: tensor<1x1x1x8xf32>) -> tensor<1x1x1x4xf32>

// CHECK-LABEL: avg_pool2d_3
// CHECK: %cst = arith.constant dense<[0, 2, 3, 1]> : tensor<4xi32>
// CHECK: %0 = "tfl.transpose"(%arg0, %cst) : (tensor<1x1x1x8xf32>, tensor<4xi32>) -> tensor<1x1x8x1xf32>

  func.func @simple_folding(%arg0: tensor<1x1x1x1xi32>, %arg1: tensor<1x1x1x1xf32>) -> tensor<?x?x?x?xf32> {
    // CHECK: %[[SHAPE:.*]] = "tf.Shape"
    // CHECK: %[[CONV:.*]] = "tf.Conv2DBackpropInput"(%[[SHAPE]]
    // CHECK-SAME: (tensor<4xi32>, tensor<1x1x1x1xf32>, tensor<1x1x1x1xf32>) -> tensor<1x1x1x1xf32>
    // CHECK: return %[[CONV]] : tensor<1x1x1x1xf32>
    %0 = "tf.Shape"(%arg0) : (tensor<1x1x1x1xi32>) -> tensor<4xi32>

func.func @testSpaceToDepthF32(%arg0: tensor<1x2x2x1xf32>) -> tensor<1x1x1x4xf32> {
  // CHECK: %[[ARG:.*]]: tensor<1x2x2x1xf32>
  // CHECK: "tfl.space_to_depth"(%[[ARG]]) <{block_size = 2 : i32}> : (tensor<1x2x2x1xf32>) -> tensor<1x1x1x4xf32>
  %0 = "tfl.space_to_depth"(%arg0) {block_size = 2: i32} : (tensor<1x2x2x1xf32>) -> tensor<1x1x1x4xf32>
  func.return %0 : tensor<1x1x1x4xf32>
}

// -----

// CHECK:   [[ZERO:%.*]] = arith.constant dense<0> : tensor<i32>
// CHECK:   [[EXP_ITEM:%.*]] = "tf.ExpandDims"([[ITEM]], [[ZERO]]) {{.*}} -> tensor<1x10xf32>
// CHECK:   [[RESULT:%.*]] = "tf.Concat"(%cst, [[INPUT]], [[EXP_ITEM]]) : {{.*}} -> tensor<?x10xf32>
// CHECK:   return [[RESULT]] : tensor<?x10xf32>
}

// -----

mlir_module = '''python
func @main(%arg0 : tensor<10xf32>, %arg1 : tensor<10xf32>) -> tensor<10x10xf32> {
   %add = "magic.op"(%arg0, %arg1) : (tensor<10xf32>, tensor<10xf32>) -> tensor<10x10xf32>
   return %ret : tensor<10x10xf32>
}
'''

@tf.function
def foo(x, y):
  return mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])