// 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>
}

// CHECK-NEXT:      %22 = mhlo.dynamic_reshape %2, %21 : (tensor<2x?x3x4xf32>, tensor<4xi32>) -> tensor<2x?x3x4xf32>
// CHECK-NEXT:      %23 = "tfl.batch_matmul"(%12, %22) <{adj_x = false, adj_y = false, asymmetric_quantize_inputs = false}> : (tensor<2x?x2x3xf32>, tensor<2x?x3x4xf32>) -> tensor<2x?x2x4xf32>
// CHECK-NEXT:      %24 = "tfl.shape"(%arg0) : (tensor<2x?x2x3xf32>) -> tensor<4xi32>

// CHECK:           %12 = "tf.Reshape"(%0, %11) : (tensor<2x?x3x4xf32>, tensor<4xi32>) -> tensor<2x?x3x4xf32>
// CHECK:           %13 = "tf.BatchMatMulV3"(%6, %12) <{adj_x = false, adj_y = false, grad_x = false, grad_y = false}> : (tensor<2x?x2x3xf32>, tensor<2x?x3x4xf32>) -> tensor<2x?x2x4xf32>
// CHECK:           %14 = "tf.Shape"(%arg0) : (tensor<2x?x2x3xf32>) -> tensor<4xi32>

  %0 = "tf.BatchMatMulV2"(%arg0, %arg1) {T = f32, adj_x = false, adj_y = false, grad_x = false, grad_y = false, device = ""} : (tensor<1x4x2xf32>, tensor<3x2x4xf32>) -> tensor<3x4x4xf32>
  func.return %0 : tensor<3x4x4xf32>
}

    %16 = "tf.Cast"(%15) {Truncate = false} : (tensor<2x2x2xi32>) -> tensor<2x2x2xf32>
    %17 = "tf.Mul"(%16, %cst_0) : (tensor<2x2x2xf32>, tensor<f32>) -> tensor<2x2x2xf32>
    %18 = "tf.AddV2"(%17, %cst) : (tensor<2x2x2xf32>, tensor<f32>) -> tensor<2x2x2xf32>
    %19 = "tf.Floor"(%18) : (tensor<2x2x2xf32>) -> tensor<2x2x2xf32>
    %20 = "tf.ClipByValue"(%19, %cst_5, %cst_6) : (tensor<2x2x2xf32>, tensor<f32>, tensor<f32>) -> tensor<2x2x2xf32>

    func.func @main() -> (tensor<1x4x4xf32>) {
      %%arg0 = "tf.Const"() {value = dense<-3.0> : tensor<1x4x2xf32>} : () -> tensor<1x4x2xf32>
      %%arg1 = "tf.Const"() {value = dense<-3.0> : tensor<1x2x4xf32>} : () -> tensor<1x2x4xf32>

      %%1 = "tf.%s"(%%arg0, %%arg1) {T = f32, adj_x = false, adj_y = false, grad_x = false, grad_y = false, device = ""} : (tensor<1x4x2xf32>, tensor<1x2x4xf32>) -> tensor<1x4x4xf32>

func.func @reshape_removeIdentity(tensor<4x4x4xf32>) -> tensor<4x4x4xf32> {
^bb0(%arg0: tensor<4x4x4xf32>) :
  %cst = arith.constant dense<[4, 4, 4]> : tensor<3xi32>
  %0 = "tfl.reshape"(%arg0, %cst) : (tensor<4x4x4xf32>, tensor<3xi32>) -> tensor<4x4x4xf32>
  func.return %0 : tensor<4x4x4xf32>

// CHECK-LABEL: func @reshape_removeIdentity
// CHECK:  return %arg0 : tensor<4x4x4xf32>
}

    %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: return %[[RESULT]]

  %indices = "tf.Const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
  %0 = "tf.DynamicStitch"(%indices, %arg0) : (tensor<2xi32>, tensor<2x2x2xf32>) -> tensor<2x2x2xf32>
  func.return %0 : tensor<2x2x2xf32>
}

// CHECK-LABEL: func @DynamicStitch_dynamic

// CHECK:           return [[RESULT]] : tensor<3x4x4xf32>
// CHECK:         }

  %0 = "tf.BatchMatMulV2"(%arg0, %arg1) {T = f32, adj_x = false, adj_y = false, device = ""} : (tensor<1x4x2xf32>, tensor<3x2x4xf32>) -> tensor<3x4x4xf32>
  func.return %0 : tensor<3x4x4xf32>
}

func.func @batchmatmulv2_lhs_batch(%arg0: tensor<3x4x2xf32>, %arg1: tensor<2x4xf32>) -> tensor<3x4x4xf32> {
// CHECK-LABEL:   func @batchmatmulv2_lhs_batch