func.func @testFullyConnectedWithBadInputShape(%arg0: tensor<2x2x11xf32>, %arg1: tensor<40x40xf32>, %arg2: none) -> tensor<40xf32> {
  // expected-error @+1 {{expect 'input' num_elements % 40 == 0, got input type 'tensor<2x2x11xf32>'}}
  %0 = "tfl.fully_connected"(%arg0, %arg1, %arg2) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<2x2x11xf32>, tensor<40x40xf32>, none) -> tensor<1x40xf32>

    : (tensor<?x3x2x1xf32>, tensor<2x1x1x1xf32>) -> tensor<?x2x2x1xf32>
  %1 = chlo.broadcast_add %0, %zp_offset : (
      tensor<?x2x2x1xf32>, tensor<?x2x2x1xf32>) -> tensor<?x2x2x1xf32>
  %2 = chlo.broadcast_add %1, %bias : (
      tensor<?x2x2x1xf32>, tensor<1xf32>) ->tensor<?x2x2x1xf32>
  return %2 : tensor<?x2x2x1xf32>
}

// -----

// CHECK-LABEL: func @dot_general_add_add

    %2 = "quantfork.stats"(%1) {layerStats = dense<[5.00000000e-6, 7.00000000e-1]> : tensor<2xf32>} : (tensor<2x2x3xf32>) -> tensor<2x2x3xf32>
    return %2 : tensor<2x2x3xf32>
  }
// CHECK: func.func private @quantize_dot_general_batch_per_tensor_quantized_fn(%[[ARG_0:.+]]: tensor<2x2x2xf32>) -> tensor<2x2x3xf32> attributes {tf._original_func_name = "main_0"}

// CHECK-NEXT:    return %2 : tensor<3x2xf32>
}

func.func @transpose_3d(%arg0: tensor<1x2x3xf32>) -> tensor<3x2x1xf32> {
  %0 = "mhlo.transpose"(%arg0) <{permutation = dense<[2, 1, 0]> : tensor<3xi64>}> : (tensor<1x2x3xf32>) -> tensor<3x2x1xf32>
  func.return %0 : tensor<3x2x1xf32>

// CHECK-LABEL:   transpose_3d
// CHECK-NEXT:    %0 = "tfl.pseudo_const"() <{value = dense<[2, 1, 0]> : tensor<3xi64>}> : () -> tensor<3xi64>

      tf_executor.fetch %outputs#1, %outputs#0 : tensor<2x2x1x2xi32>, tensor<2x2x1x2xf32>
    }
    func.return %0#0, %0#1 : tensor<2x2x1x2xi32>, tensor<2x2x1x2xf32>
  }
}

// CHECK:        name: "serving_default_x"
// CHECK-NEXT:   op: "_Arg"

func.func @eliminateExtraSelectLhs(%arg0: tensor<4x2x1xf32>, %arg1: tensor<4x2x1xi1>) -> (tensor<4x2x1xf32>) {
  %cst0 = arith.constant dense<1.0> : tensor<2x2xf32>
  %cst1 = arith.constant dense<2.0> : tensor<2xf32>
  %cst_zeros = arith.constant dense<0.0> : tensor<4x2x1xf32>

  %0 = "tfl.select_v2"(%arg1, %cst_zeros, %arg0) : (tensor<4x2x1xi1>, tensor<4x2x1xf32>, tensor<4x2x1xf32>) -> tensor<4x2x1xf32>

// 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>

}

// -----

func.func @trivial_dynamic_update_slice(%arg0: tensor<2x7x14xf32>, %arg1: tensor<2x7x14xf32>) -> tensor<2x7x14xf32> {
  %0 = arith.constant dense<0> : tensor<3xi32>
  %1 = "tfl.dynamic_update_slice"(%arg0, %arg1, %0) : (tensor<2x7x14xf32>, tensor<2x7x14xf32>, tensor<3xi32>) -> tensor<2x7x14xf32>
  // CHECK: return %arg1
  func.return %1 : tensor<2x7x14xf32>
}

// -----

  module {
    func.func @main(%arg0: tensor<2x2x2xf32>, %arg1: tensor<2x2x2xf32>) -> tensor<2x2x2xf32> attributes {_from_xla_call_module} {
      %0 = stablehlo.dot_general %arg0, %arg1,
        batching_dims = [0] x [0],
        contracting_dims = [2] x [1],
        precision = [DEFAULT, DEFAULT]
      : (tensor<2x2x2xf32>, tensor<2x2x2xf32>) -> tensor<2x2x2xf32>
        return %0 : tensor<2x2x2xf32>
    }
  }
)mlir";

func.func @Batchmatmul2FullyconnectedBatchedY(%arg0: tensor<4x128x2xf32>) -> (tensor<4x128x1xf32>) {
  %0 = arith.constant dense<42.> : tensor<4x2x1xf32>
  %1 = "tfl.batch_matmul"(%arg0, %0) {adj_x = false, adj_y = false, asymmetric_quantize_inputs = false} : (tensor<4x128x2xf32>, tensor<4x2x1xf32>) -> tensor<4x128x1xf32>
  func.return %1 : tensor<4x128x1xf32>

  // CHECK: %[[BMM_RES:.*]] = "tfl.batch_matmul"
  // CHECK-NEXT: return %[[BMM_RES]]