// CHECK-SAME:                         %[[VAL_0:.*]]: tensor<1x8x8x207xf32>,
// CHECK-SAME:                         %[[VAL_1:.*]]: tensor<3x3x207x16xf32>) -> tensor<1x8x8x16xf32> {
// CHECK:           %[[VAL_2:.*]] = "tf.Conv2D"(%[[VAL_0]], %[[VAL_1]]) <{data_format = "NHWC", dilations = [1, 1, 1, 1], explicit_paddings = [], padding = "SAME", strides = [1, 1, 1, 1], use_cudnn_on_gpu = true}> : (tensor<1x8x8x207xf32>, tensor<3x3x207x16xf32>) -> tensor<1x8x8x16xf32>

  %0:4 = "tfl.split"(%arg0, %arg1) {num_splits = 4 : i32, tac.device = "CPU"} : (tensor<i32>, tensor<1x8x8x1024xf32>) -> (tensor<1x8x8x256xf32>, tensor<1x8x8x256xf32>, tensor<1x8x8x256xf32>, tensor<1x8x8x256xf32>)
  func.return %0#0, %0#1, %0#3 : tensor<1x8x8x256xf32>, tensor<1x8x8x256xf32>, tensor<1x8x8x256xf32>
}

  func.return %0 : tensor<16x8x8x1xf32>
  }
}

// CHECK: module {
// CHECK-NEXT:   func.func @main(%arg0: tensor<8x8x1x207xf32>, %arg1: tensor<3x3x16x207xf32>) -> tensor<16x8x8x1xf32> {

  %1 = "tfl.mul"(%0, %cst_2) {fused_activation_function = "NONE"} : (tensor<1x8x8x256xf32>, tensor<1x1x1x256xf32>) -> tensor<1x8x8x256xf32>
  func.return %1 : tensor<1x8x8x256xf32>

// CHECK-DAG: %[[CST_0:.*]] = arith.constant dense<2.800000e+00> : tensor<256x3x3x207xf32>