//
//
//   SpaceToBatchND -> Expand -> Conv2D -> Squeeze -> BatchToSpaceND -> BiasAdd
//
//   SpaceToBatchND -> Expand -> Conv2D -> Squeeze -> Pad -> BatchToSpaceND ->
//   BiasAdd
//
//   SpaceToBatchND -> Expand -> Conv2D -> Squeeze -> BiasAdd -> BatchToSpaceND
//
//   SpaceToBatchND -> Conv2D -> Pad -> BatchToSpaceND -> BiasAdd
//
//   SpaceToBatchND -> Conv2D -> BatchToSpaceND -> BiasAdd
//
//

Michael Levesque-Dion <******@****.***> 1706075999 -0800

    %zp_offset: tensor<?x2x2x1xi32>, %bias: tensor<1xi32>
  ) -> tensor<?x2x2x1xi32> {
  // CHECK-DAG: %[[conv:.*]] = mhlo.convolution
  // CHECK-DAG: %[[combined:.*]] = chlo.broadcast_add %[[zp_offset:.*]], %[[bias:.*]]
  // CHECK-DAG: %[[result:.*]] = chlo.broadcast_add %[[conv]], %[[combined]]
  // CHECK: return %[[result]]
  %0 = mhlo.convolution(%lhs, %rhs)
      dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f],

// CHECK:           %[[CONV:.*]] = "tf.Conv2D"(%[[SLICED_ARG0]], %[[ARG1]])
// CHECK-SAME:      explicit_paddings = [0, 0, 4, 0, 0, 2, 0, 0]
// CHECK-SAME:      (tensor<128x5x4x64xf32>, tensor<3x2x64x4xf32>) -> tensor<128x4x3x4xf32>
// CHECK:           return %[[CONV]] : tensor<128x4x3x4xf32>
// CHECK:         }

  %b = arith.constant dense<-1.23697901> : tensor<64xf32>
  %conv = "tfl.conv_2d"(%arg0, %w, %b) {dilation_h_factor = 1 : i32, dilation_w_factor = 1 : i32, fused_activation_function = "NONE", padding = "SAME", stride_h = 2 : i32, stride_w = 2 : i32} : (tensor<1x224x224x3xf32>, tensor<64x3x3x3xf32>, tensor<64xf32>) -> tensor<1x112x112x64xf32>
  func.return %conv : tensor<1x112x112x64xf32>

  }
  func.func private @composite_conv2d_with_bias_and_relu6_fn_1(%arg0: tensor<*xf32>, %arg1: tensor<*xf32>, %arg2: tensor<2xf32>) -> tensor<*xf32> attributes {tf_quant.composite_function} {

  %cst = "tf.Const"() {value = dense<0.000000e+00> : tensor<2xf32>} : () -> tensor<2xf32>
  %cst_1 = "tf.Const"() {value = dense<3.000000e+00> : tensor<2x3x3x2xf32>} : () -> tensor<2x3x3x2xf32>
  %0 = "tf.Conv2D"(%arg0, %cst_1) {
    data_format = "NHWC", device = "", dilations = [1, 1, 1, 1], explicit_paddings = [],
    padding = "SAME", strides = [1, 1, 2, 1], use_cudnn_on_gpu = true

		return mkcall("countrunes", n.Type(), init, typecheck.Conv(n.X.(*ir.ConvExpr).X, types.Types[types.TSTRING]))
	}
	if isByteCount(n) {
		conv := n.X.(*ir.ConvExpr)
		walkStmtList(conv.Init())
		init.Append(ir.TakeInit(conv)...)
		_, len := backingArrayPtrLen(cheapExpr(conv.X, init))
		return len
	}
	if isChanLenCap(n) {
		name := "chanlen"
		if n.Op() == ir.OCAP {

  // CHECK: %[[ARG_PERM:.*]] = "tf.Const"() <{value = dense<[0, 2, 3, 1]> : tensor<4xi64>}>
  // CHECK: %[[ARG_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%arg0, %[[ARG_PERM]])

  // CHECK: %[[CONV2D:[0-9]*]] = "tf.Conv2D"(%[[ARG_TRANSPOSE]], %arg1)
  // CHECK-SAME: data_format = "NHWC"
  // CHECK-SAME: dilations = [1, 3, 4, 2]
  // CHECK-SAME: explicit_paddings = [1, 2, 5, 6, 7, 8, 3, 4]
  // CHECK-SAME: padding = "EXPLICIT"

  %0 = "tf.Conv2D"(%arg0, %arg1) {
    data_format = "NHWC", device = "", dilations = [1, 1, 1, 1], explicit_paddings = [],
    padding = "SAME", strides = [1, 1, 2, 1], use_cudnn_on_gpu = true
  } : (tensor<1x3x4x3xf32>, tensor<2x3x3x2xf32>) -> tensor<*xf32>
  %1 = "tf.Relu6"(%0) {device = ""} : (tensor<*xf32>) -> tensor<*xf32>


  %3 = "tf.Conv2D"(%arg0, %arg1) {