// CHECK: %[[CONST_1:.*]] = arith.constant dense
// CHECK-NOT: tensor<2x3x3x2xf32>
// CHECK-SAME: tensor<2x3x1x6xf32>
// CHECK: %[[PARTITIONEDCALL_0:.*]] = "tf.PartitionedCall"(%arg0, %[[CONST_1:.*]]) <{config = "", config_proto = "", executor_type = "", f = @composite_depthwise_conv2d_fn_0}> {_tfl_quant_trait = "fully_quantizable"} : (tensor<1x3x4x3xf32>, tensor<2x3x1x6xf32>) -> tensor<*xf32>

func.func @tfInplaceUpdate(%arg0: tensor<2x1x2xf32>) -> tensor<2x1x2xf32> {
  %1 = arith.constant dense<1> : tensor<1xi32>
  %2 = arith.constant dense<2.0> : tensor<1x1x2xf32>
  %3 = "tf.InplaceUpdate"(%arg0, %1, %2) {device = ""}
    : (tensor<2x1x2xf32>, tensor<1xi32>, tensor<1x1x2xf32>) -> tensor<2x1x2xf32>
  func.return %3 : tensor<2x1x2xf32>
}

}

// CHECK-LABEL: @tfInplaceUpdate

func.func @my_add_n_single_input(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
  %0 = "tf.MyAddN"(%arg0) {N=1:i32} : (tensor<2x3xf32>) -> tensor<2x3xf32>
  func.return %0 : tensor<2x3xf32>

// CHECK-NEXT: return %arg0 : tensor<2x3xf32>
}

// CHECK-LABEL: my_add_n_multiple_inputs
func.func @my_add_n_multiple_inputs(%arg0: tensor<2x3xf32>, %arg1: tensor<2x3xf32>, %arg2: tensor<2x3xf32>) -> tensor<2x3xf32> {

  // CHECK-LABEL: test_batch_matmul_broadcast_to_einsum
  // CHECK: "tf.Einsum"(%arg0, %arg1) <{equation = "...mk,...kn->...mn"}> : (tensor<2x2x4xf32>, tensor<2x4x2xf32>) -> tensor<2x2x2xf32>
  %0 = "tf.BatchMatMul"(%arg0, %arg1) {adj_x = false, adj_y = false} : (tensor<2x2x4xf32>, tensor<2x4x2xf32>) -> tensor<2x2x2xf32>
  func.return %0: tensor<2x2x2xf32>
}

  } : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>
  return %0 : tensor<1x3xf32>
}

// CHECK-LABEL: func.func @qdq_for_dot_general_weight_empty
// CHECK-SAME: (%[[ARG_0:.+]]: tensor<1x2xf32>) -> tensor<1x3xf32>
// CHECK: %[[CST:.+]] = "tf.Const"() <{value = dense<3.000000e-01> : tensor<2x3xf32>}> : () -> tensor<2x3xf32>

  %1 = "quantfork.stats"(%0) {bitsNum = 8 : i64, layerStats = dense<[-2.0, 2.0]> : tensor<2xf32>, narrowRange = false} : (tensor<2x3xf32>) -> tensor<2x3xf32>
  %2 = stablehlo.convert %1 : (tensor<2x3xf32>) -> (tensor<2x3xf32>)
  func.return %2 : tensor<2x3xf32>
}

// -----

// CHECK-LABEL: func @dot_redundant_stats

    %3 = stablehlo.concatenate %2, %1, dim = 0 : (tensor<1x3xf32>, tensor<1x3xf32>) -> tensor<2x3xf32>
    return %3 : tensor<2x3xf32>
  }
  func.func private @composite_dot_general_fn_1(%arg0: tensor<1x2xf32>, %arg1: tensor<2x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module, tf_quant.composite_function} {

// CHECK-NEXT:}

func.func @transpose(%arg0: tensor<2x3x2xi32>) -> tensor<2x3x2xi32> {
  %0 = "vhlo.transpose_v1"(%arg0) <{permutation = #vhlo.tensor_v1<dense<[2, 1, 0]> : tensor<3xi64>>}> : (tensor<2x3x2xi32>) -> tensor<2x3x2xi32>
  return %0 : tensor<2x3x2xi32>
}

// CHECK:func.func private @transpose(%arg0: tensor<2x3x2xi32>) -> tensor<2x3x2xi32> {

func.func @transpose_simple_2d() -> tensor<3x2xf32> {
  %0 = stablehlo.constant dense<[[0.000000e+0, 1.000000e+0, 2.000000e+0], [3.000000e+0, 4.000000e+0, 5.000000e+0]]> : tensor<2x3xf32>
  %1 = stablehlo.transpose %0, dims = [1, 0] : (tensor<2x3xf32>) -> tensor<3x2xf32>
  return %1 : tensor<3x2xf32>
}

  // 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