func.func @StridedSliceEllipsisMaskAfter(%arg0: tensor<21x15x7xf32>) -> tensor<5x15x7xf32> {
  %cst = arith.constant dense<0> : tensor<2xi32>
  %cst_0 = arith.constant dense<1> : tensor<2xi32>
  %0 = "tf.StridedSlice"(%arg0, %cst, %cst, %cst_0) {begin_mask = 0 : i64, ellipsis_mask = 2 : i64, end_mask = 0 : i64, new_axis_mask = 0 : i64, shrink_axis_mask = 0 : i64} : (tensor<21x15x7xf32>, tensor<2xi32>, tensor<2xi32>, tensor<2xi32>) -> tensor<5x15x7xf32>

  // CHECK: return %[[v3:.*]] : tensor<5x7xf32>
}

func.func @einsum_transpose_matmul(%arg0: tensor<2x5x7xf32>, %arg1: tensor<5x3x2xf32>) -> tensor<5x3x7xf32> {
  %0 = "tf.Einsum"(%arg0, %arg1) {T = "tfdtype$DT_FLOAT", equation = "lbh,bkl->bkh"}: (tensor<2x5x7xf32>, tensor<5x3x2xf32>) -> tensor<5x3x7xf32>
  func.return %0 : tensor<5x3x7xf32>
  // CHECK-LABEL: einsum_transpose_matmul

// CHECK: return %[[RESULT0]]
}

// CHECK-LABEL: testMatmulAsymAttributeTrue
func.func @testMatmulAsymAttributeTrue(%arg0: tensor<4x10x15xf32>, %arg1: tensor<4x15x17xf32>) -> tensor<4x10x17xf32> {
  %0 = "tfl.batch_matmul"(%arg0, %arg1) {adj_x = false, adj_y = false, asymmetric_quantize_inputs = true} : (tensor<4x10x15xf32>, tensor<4x15x17xf32>) -> tensor<4x10x17xf32>

func.func @broadcast_mul_implicit_no_fold(%arg0: tensor<5x7xf32>, %arg1: tensor<5xf32>) -> tensor<3x5x7xf32> {
  %cst = arith.constant dense<[3, 5, 7]> : tensor<3xi32>
  %0 = "tf.BroadcastTo"(%arg1, %cst) : (tensor<5xf32>, tensor<3xi32>) -> tensor<3x5x7xf32>
  %1 = "tf.Mul"(%arg0, %0) : (tensor<5x7xf32>, tensor<3x5x7xf32>) -> tensor<3x5x7xf32>
  func.return %1 : tensor<3x5x7xf32>
  // CHECK: %[[C0:.*]] = arith.constant dense<[3, 5, 7]> : tensor<3xi32>

# CHECK:           %[[VAL_12:.*]] = "tfl.pack"(%[[VAL_10]], %[[VAL_11]]) <{axis = 0 : i32, values_count = 2 : i32}> : (tensor<5x7xf32>, tensor<5x7xf32>) -> tensor<2x5x7xf32>
# CHECK:           return %[[VAL_12]] : tensor<2x5x7xf32>

# CHECK:     %[[VAL_2:.*]] = "tfl.batch_matmul"(%[[VAL_0]], %[[VAL_1]]) <{adj_x = false, adj_y = false}> : (tensor<2x5x3xf32>, tensor<3x7xf32>) -> tensor<2x5x7xf32>
# CHECK:     return %[[VAL_2]] : tensor<2x5x7xf32>

    func.func @main(%a : tensor<5x14x1xf32>, %b : tensor<1x14x32xf32>) -> tensor<?x?x?xf32> attributes {tf.entry_function = {control_outputs = "", inputs = "i,j", outputs = "identity_RetVal"}} {
      %c = "mhlo.einsum"(%a, %b) {einsum_config = "bji,bjk->bik"} : (tensor<5x14x1xf32>, tensor<1x14x32xf32>) -> tensor<?x?x?xf32>
      return %c : tensor<?x?x?xf32>
    }
  })";

  // CHECK: "tf.BatchMatMulV2"(%arg0, %arg1) <{adj_x = false, adj_y = false, grad_x = false, grad_y = false}> {device = "/job:localhost/replica:0/task:0/device:GPU:0"}
  %0 = "tf.BatchMatMul"(%arg0, %arg1) <{adj_x = false, adj_y = false}> {device = "/job:localhost/replica:0/task:0/device:GPU:0"} : (tensor<2x3x5xf32>, tensor<2x5x7xf32>) -> tensor<2x3x7xf32>

}

func.func @strided_slice_big_dims(%arg0: tensor<5x6x7xf32>, %arg1: tensor<3xi32>, %arg2: tensor<3xi32>, %arg3: tensor<3xi32>) -> tensor<1x1x5x6x7xf32> {
  %0 = "tf.StridedSlice"(%arg0, %arg1, %arg2, %arg3) {begin_mask = 0 : i64, ellipsis_mask = 0 : i64, end_mask = 0 : i64, new_axis_mask = 7 : i64, shrink_axis_mask = 0 : i64, offset = false} : (tensor<5x6x7xf32>, tensor<3xi32>, tensor<3xi32>, tensor<3xi32>) -> tensor<1x1x5x6x7xf32>

func.func @testBatchMatMul(%lhs: tensor<2x?x2x?x3x5xf32>, %rhs: tensor<2x2x?x?x5x7xf32>) {
  %0 = "tf.BatchMatMul"(%lhs, %rhs) : (tensor<2x?x2x?x3x5xf32>, tensor<2x2x?x?x5x7xf32>) -> tensor<2x?x?x?x3x7xf32>
  func.return
}

// -----

// Mismatching batch dimensions.
func.func @testBatchMatMul(%lhs: tensor<1x3x5xf32>, %rhs: tensor<2x5x7xf32>) {