func.return %2 : tensor<4x4x4xf32>
}

      %1 = "tf.XlaCallModule"(%0, %cst_0, %cst_1) <{Sout = [#tf_type.shape<1x4x4x3>], module = "", version = 9 : i64}> {_entry_function = @composite_fn_2, _original_entry_function = "composite_fn_2", _tfl_quant_trait = "fully_quantizable"} : (tensor<1x4x4x3xf32>, tensor<3x1x1x3xf32>, tensor<3xf32>) -> tensor<1x4x4x3xf32>
      return %1 : tensor<1x4x4x3xf32>

  func.return %17 : tensor<1x2x3xf32>

  // CHECK: %[[NONE:.*]] = "tfl.no_value"() <{value}> : () -> none
  // CHECK: %[[DQ_1:.*]] = "tfl.dequantize"({{.*}}) : (tensor<1x1xf16>) -> tensor<1x1xf32>
  // CHECK: %[[DQ_2:.*]] = "tfl.dequantize"({{.*}}) : (tensor<1x1xf16>) -> tensor<1x1xf32>
  // CHECK: %[[DQ_3:.*]] = "tfl.dequantize"({{.*}}) : (tensor<1x1xf16>) -> tensor<1x1xf32>

func.func @batchmatmulv2_basic(%arg0: tensor<1x4x2xf32>, %arg1: tensor<3x2x4xf32>) -> tensor<3x4x4xf32> {
// CHECK-LABEL:   func @batchmatmulv2_basic
// CHECK-SAME:        ([[LHS:%.*]]: tensor<1x4x2xf32>, [[RHS:%.*]]: tensor<3x2x4xf32>) -> tensor<3x4x4xf32>
// CHECK:           [[LHSSHAPE:%.*]] = shape.shape_of [[LHS]] : tensor<1x4x2xf32>
// CHECK:           [[RHSSHAPE:%.*]] = shape.shape_of [[RHS]] : tensor<3x2x4xf32>

  func.return %0: tensor<1x2x4xf32>
}

func.func @test_batch_matmul_adj_to_einsum(%arg0: tensor<1x2x3xf32>, %arg1: tensor<4x3xf32>) -> tensor<1x2x4xf32> {
  // CHECK-LABEL: test_batch_matmul_adj_to_einsum
  // CHECK: %[[RES_EINSUM:[0-9]*]] = "tf.Einsum"(%arg0, %arg1) <{equation = "...mk,...nk->...mn"}> : (tensor<1x2x3xf32>, tensor<4x3xf32>) -> tensor<1x2x4xf32>
  // CHECK: return %[[RES_EINSUM]] : tensor<1x2x4xf32>

  }
  tfr.return %res : !tfr.tensor
}

// CHECK-LABEL: pack_one
func.func @pack_one(%arg0: tensor<2x3xf32>) -> tensor<1x2x3xf32> {
  %0 = "tf.MyPack"(%arg0) {N=1:i32, axis=0:i32} : (tensor<2x3xf32>) -> tensor<1x2x3xf32>
  func.return %0 : tensor<1x2x3xf32>

// CHECK-NEXT: %[[AXIS:.*]] = arith.constant 0 : i32
// CHECK-NEXT: %[[CAST:.*]] = "tfr.cast"(%arg0) : (tensor<2x3xf32>) -> !tfr.tensor

    %0 = stablehlo.constant dense<1.000000e+00> : tensor<4x3xf32>
    %1 = "quantfork.qcast"(%0) {volatile} : (tensor<4x3xf32>) -> tensor<4x3x!quant.uniform<i8<-127:127>:f32:1, {5.000000e-03, 5.000000e-03, 5.000000e-03}>>
    %2 = "quantfork.dcast"(%1) : (tensor<4x3x!quant.uniform<i8<-127:127>:f32:1, {5.000000e-03, 5.000000e-03, 5.000000e-03}>>) -> tensor<4x3xf32>

  %0 = stablehlo.constant dense<1.0> : tensor<4x3xf32>
  %1 = stablehlo.dot_general %arg0, %0, contracting_dims = [1] x [0], precision = [DEFAULT, DEFAULT] : (tensor<1x4xf32>, tensor<4x3xf32>) -> tensor<1x3xf32>
  return %1 : tensor<1x3xf32>
}
// CHECK: @main(%[[ARG_0:.+]]: tensor<1x4xf32>) -> tensor<1x3xf32>
// CHECK-DAG: %[[CST:.+]] = stablehlo.constant dense<1.000000e+00> : tensor<4x3xf32>

  %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>
}