func.func @ReorderAddWithConstant(%arg0: tensor<2x2xf32>) -> tensor<2x2xf32> {
  %cst = arith.constant dense<1.0> : tensor<2x2xf32>
  %cst_1 = arith.constant dense<2.0> : tensor<2x2xf32>
  %0 = "tfl.add"(%arg0, %cst) {fused_activation_function = "NONE"} : (tensor<2x2xf32>, tensor<2x2xf32>) -> tensor<2x2xf32>
  %1 = "tfl.add"(%0, %cst_1) {fused_activation_function = "NONE"} : (tensor<2x2xf32>, tensor<2x2xf32>) -> tensor<2x2xf32>

    %7 = stablehlo.slice %6 [1:3, 2:4] : (tensor<3x4xf32>) -> tensor<2x2xf32>
    %8 = "quantfork.qcast"(%7) {volatile} : (tensor<2x2xf32>) -> tensor<2x2x!quant.uniform<i8:f32, 0.13170163023705575:-1>>
    %9 = "quantfork.dcast"(%8) : (tensor<2x2x!quant.uniform<i8:f32, 0.13170163023705575:-1>>) -> tensor<2x2xf32>
    return %9 : tensor<2x2xf32>
  }

  // CHECK: quantized_dot_general_fn_1

  // CHECK: %[[v0:.*]] = "tf.Transpose"(%arg0, %[[cst]]) : (tensor<2x5x7xf32>, tensor<3xi32>) -> tensor<5x7x2xf32>
  // CHECK: %[[v1:.*]] = "tf.Transpose"(%arg1, %[[cst_0]]) : (tensor<5x3x2xf32>, tensor<3xi32>) -> tensor<5x2x3xf32>
  // CHECK: %[[v2:.*]] = "tf.BatchMatMulV2"(%[[v0]], %[[v1]]) <{adj_x = false, adj_y = false}> : (tensor<5x7x2xf32>, tensor<5x2x3xf32>) -> tensor<5x7x3xf32>

}

// CHECK-LABEL: QuantizeConcat
func.func @QuantizeConcat(tensor<1x2xf32>, tensor<1x2xf32>) -> tensor<2x2x!quant.uniform<u8:f32, 1.000000e-01:128>> {
^bb0(%arg0: tensor<1x2xf32>, %arg1: tensor<1x2xf32>):
  %0 = "tfl.concatenation"(%arg0, %arg1) {axis = 0 : i32, fused_activation_function = "NONE"} : (tensor<1x2xf32>, tensor<1x2xf32>) -> tensor<2x2xf32>

  %w = arith.constant dense<127.0> : tensor<512x12xf32>
  %b = arith.constant dense<0.0> : tensor<512xf32>
  %fc = "tfl.fully_connected"(%arg0, %w, %b) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<1x224x224x3xf32>, tensor<512x12xf32>, tensor<512xf32>) -> tensor<1x112x112x512xf32>
  func.return %fc : tensor<1x112x112x512xf32>

  %0 = "tf_device.cluster_func"(%arg0, %arg1) {func = @_func, use_spmd_for_xla_partitioning = true, use_tpu = true, num_cores_per_replica = 2 : i64} : (tensor<2x4xf32>, tensor<4x2xf32>) -> tensor<2x2xf32>
  %1:2 = "tf.TPUPartitionedOutputV2"(%0) {device = "", partition_dims = [2, 1]} : (tensor<2x2xf32>) -> (tensor<1x2xf32>, tensor<1x2xf32>)

    return %2 : tensor<1x2xf32>
  }

  func.func private @composite_add_fn(%arg0: tensor<1x2xf32>, %arg1: tensor<1x2xf32>) -> tensor<1x2xf32> attributes {_from_xla_call_module} {
    %0 = stablehlo.add %arg0, %arg1 : tensor<1x2xf32>
    %1 = stablehlo.add %0, %arg1 : tensor<1x2xf32>
    return %1 : tensor<1x2xf32>
  }
}

// -----

  %0 = "tf.Div"(%arg0, %cst) : (tensor<2x2xf32>, tensor<2x2xf32>) -> tensor<2x2xf32>
  func.return %0 : tensor<2x2xf32>

  // CHECK-LABEL: RemoveTrivialDiv
  // CHECK-NEXT: return %arg0 : tensor<2x2xf32>
}

func.func @RemoveTrivialRealDiv(%arg0: tensor<2x2xf32>, %arg1: tensor<2x2xf32>) -> tensor<2x2xf32> {
  %cst = arith.constant dense<1.0> : tensor<2x2xf32>

    has_token_input_output = false, module = "", platforms = [],
    version = 5 : i64
  } : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>
  return %0 : tensor<1x3xf32>
}

// CHECK-LABEL: func.func @qdq_for_dot_general_weight_empty
// CHECK-SAME: (%[[ARG_0:.+]]: tensor<1x2xf32>) -> tensor<1x3xf32>
// CHECK: %[[CST:.+]] = "tf.Const"() <{value = dense<3.000000e-01> : tensor<2x3xf32>}> : () -> tensor<2x3xf32>

    %1 = "tf.PartitionedCall"(%arg0, %cst_0) {_tfl_quant_trait = "fully_quantizable", config = "", config_proto = "", executor_type = "", f = @composite_matmul_fn_1} : (tensor<2x12xf32>, tensor<12x2xf32>) -> tensor<*xf32>
    func.return %1: tensor<*xf32>
  }