// CHECK: %[[v2:.*]] = "tf.Transpose"(%[[v1]], %[[cst]]) : (tensor<2x5x7x13xf32>, tensor<4xi32>) -> tensor<2x7x5x13xf32>
  // CHECK: return %[[v2]] : tensor<2x7x5x13xf32>
}

func.func @einsum_fourdtransposeall(%arg0: tensor<2x5x7x3xf32>, %arg1: tensor<2x11x7x3xf32>) -> tensor<2x7x11x5xf32> {

    %12 = "quantfork.qcast"(%11) {volatile} : (tensor<2x3x1x3xf32>) -> tensor<2x3x1x3x!quant.uniform<i8:f32, 3.000000e-01:1>>
    %13 = "quantfork.dcast"(%12) : (tensor<2x3x1x3x!quant.uniform<i8:f32, 3.000000e-01:1>>) -> tensor<2x3x1x3xf32>
    return %13 : tensor<2x3x1x3xf32>
  }

  // CHECK: quantized_dot_general_fn_1

  %0 = stablehlo.constant dense<2.000000e+00> : tensor<1x4x3x3xf32>
  %1 = stablehlo.transpose %arg0, dims = [0, 3, 1, 2] : (tensor<1x3x3x4xf32>) -> tensor<1x4x3x3xf32>
  %2 = stablehlo.add %1, %0 : tensor<1x4x3x3xf32>
  return %2 : tensor<1x4x3x3xf32>
}
// CHECK-SAME: (%[[ARG_0:.+]]: tensor<1x3x3x4xf32>) -> tensor<1x4x3x3xf32>
// CHECK-DAG: %[[CONST_0:.+]] = stablehlo.constant

func.func @CheckNumericVerifyWholeModel(%arg0: tensor<1x4x4x3xf32>) -> tensor<1x1x1x3xf32> {
  %0 = "tfl.quantize"(%arg0) {qtype = tensor<1x4x4x3x!quant.uniform<i8:f32, 0.1>>, volatile} : (tensor<1x4x4x3xf32>) -> tensor<1x4x4x3x!quant.uniform<i8:f32, 0.1>>
  %1 = "tfl.dequantize"(%0) : (tensor<1x4x4x3x!quant.uniform<i8:f32, 0.1>>) -> tensor<1x4x4x3xf32>

  %cst = "tf.Const"() {value = dense<1.000000e+00> : tensor<2x3x3x3xf32>} : () -> tensor<2x3x3x3xf32>
  %cst_0 = "tf.Const"() {value = dense<0.500000e+00> : tensor<1x3x2x3xf32>} : () -> tensor<1x3x2x3xf32>
  %0 = "tf.Conv2D"(%arg0, %cst) {data_format = "NHWC", dilations = [1, 1, 2, 1], explicit_paddings = [], padding = "SAME", strides = [1, 1, 2, 1], use_cudnn_on_gpu = true} : (tensor<1x3x4x3xf32>, tensor<2x3x3x3xf32>) -> tensor<1x3x2x3xf32>

    func.return %result : tensor<10x24x24x64xf32>
  }

  // CHECK-LABEL: @max_pool_grad_same
  func.func @max_pool_grad_same(%orig_input: tensor<2x13x25x7xf32>, %orig_output: tensor<2x4x7x7xf32>, %grad: tensor<2x4x7x7xf32>) -> tensor<2x13x25x7xf32> {
    // CHECK: padding = dense<{{\[\[}}0, 0], [0, 1], [1, 1], [0, 0]]> : tensor<4x2xi64>
    %result = "tf.MaxPoolGrad"(%orig_input, %orig_output, %grad) {
      data_format = "NHWC",

// CHECK:           %[[VAL_6:.*]] = arith.constant dense<[0, 3, 1, 2]> : tensor<4xi32>
// CHECK:           %[[VAL_7:.*]] = "tfl.transpose"(%[[VAL_5]], %[[VAL_6]]) : (tensor<1x4x4x3xf32>, tensor<4xi32>) -> tensor<1x3x4x4xf32>