// CHECK-LABEL: func @testWhileCasts(%arg0: tensor<!tf_type.variant<tensor<1x3xf32>>>) -> tensor<!tf_type.variant<tensor<*xf32>>>
func.func @testWhileCasts(%arg0: tensor<!tf_type.variant<tensor<1x3xf32>>>) -> (tensor<!tf_type.variant<tensor<*xf32>>>) {
  %0 = "tf.While"(%arg0) {
    cond = @testWhileCond, body = @testWhileBody, is_stateless = false
  } : (tensor<!tf_type.variant<tensor<1x3xf32>>>) -> (tensor<!tf_type.variant<tensor<*xf32>>>)

      %0 = stablehlo.dot_general %arg0, %arg1, contracting_dims = [1] x [0], precision = [DEFAULT, DEFAULT] : (tensor<1x1024xf32>, tensor<1024x3xf32>) -> tensor<1x3xf32>
      return %0 : tensor<1x3xf32>
    }
  }
)mlir";

TEST_F(LiftAsFunctionCallTest, LiftedFunctionSucceeds) {

  // CHECK-NEXT:     %[[GROUP:.*]] = "tf.Const"()
  // CHECK-SAME:       [0, 1, 2, 3]
  // CHECK-NEXT:     %[[REDUCED:.*]] = "tf.XlaAllReduce"(%[[ID]], %[[GROUP]]) <{mode = "CrossReplica", reduce_op = "Add"}> : (tensor<f32>, tensor<1x4xi32>) -> tensor<f32>
  // CHECK-NEXT:     "tf.OpA"(%[[REDUCED]]) : (tensor<f32>) -> ()
  tf_device.replicate {n = 4 : i32} {
    "tf_device.cluster"() ({
      "tf.OpA"(%arg0) : (tensor<f32>) -> ()
      tf_device.return

          }
        }
        tensor_content: "\x01\x00\x00\x00\x03\x00\x00\x00\x02\x00\x00\x00\x05\x00\x00\x00\x04\x00\x00\x00\x07\x00\x00\x00"
        # CHECK: value = dense<{{\[\[}}1, 3, 2], [5, 4, 7]]> : tensor<2x3xi32>
      }
    }
  }

  %group_assignment = "tf.Const"() { value = dense<[[0, 1]]> : tensor<1x2xi32> } : () -> tensor<1x2xi32>
  %instance_key = "tf.Const"() { value = dense<3> : tensor<i32> } : () -> tensor<i32>
  %group_size, %group_key = "tf.CollectiveAssignGroupV2"(%group_assignment, %rank, %key_base) {} : (tensor<1x2xi32>, tensor<i32>, tensor<i32>) -> (tensor<i32>, tensor<i32>)
  // CHECK-NOT: "tf.CollectiveAssignGroupV2"

        }) : () -> tensor<f32>
        "tf_device.cluster"() ({
          %group = "tf.Const"() <{value = dense<[[0, 1]]> : tensor<1x2xi32>}> : () -> tensor<1x2xi32>
          %arg1_reduced = "tf.XlaAllReduce"(%arg1_id, %group) <{mode = "CrossReplica", reduce_op = "Add"}> : (tensor<f32>, tensor<1x2xi32>) -> tensor<f32>
          "tf.OpA"(%arg1_reduced) : (tensor<f32>) -> ()
          tf_device.return
        }) : () -> ()

func.func @dot_general_with_bias_same_shape_fn(%arg0: tensor<1x2xf32>) -> tensor<1x3xf32> {
  %0 = stablehlo.constant dense<2.000000e+00> : tensor<2x3xf32>
  %1 = stablehlo.constant dense<2.000000e+00> : tensor<1x3xf32>
  %2 = stablehlo.dot_general %arg0, %0, contracting_dims = [1] x [0], precision = [DEFAULT, DEFAULT] : (tensor<1x2xf32>, tensor<2x3xf32>) -> tensor<1x3xf32>

  %0 = "tfl.quantize"(%arg0) {qtype = tensor<1x3x3x!quant.uniform<i8:f32, 0.003:-128>>} : (tensor<1x3x3xf32>) -> tensor<1x3x3x!quant.uniform<i8:f32, 0.003:-128>>
  %1 = "tfl.logistic"(%0) : (tensor<1x3x3x!quant.uniform<i8:f32, 0.003:-128>>) -> tensor<1x3x3x!quant.uniform<i8:f32, 3.906250e-03:-128>>

  %one = "glue.constant"() { value = 1: i32 } : () -> i32
  %done = "glue.compare" (%one, %one) { predicate = #glue<"compare LTE"> } : (i32, i32) -> i1
  %2 = mhlo.constant dense<[[1.1]]> : tensor<1x1xf32>
  %3 = mhlo.multiply %2, %2 : tensor<1x1xf32>
  %cst = "tf.Const"() {value = dense<0.0> : tensor<f32>} : () -> tensor<f32>
  %0 = "tf.AddV2"(%arg0, %cst) {device = "/device:CPU:0"} : (tensor<f32>, tensor<f32>) -> tensor<f32>

  %1 = "tf.ReadVariableOp"(%0) {device = "/CPU:0"} : (tensor<!tf_type.resource<tensor<1x1xf32>>>) -> tensor<1x1xf32>
  %2:2 = "tf.Split"(%cst, %arg0) {device = "/CPU:0"} : (tensor<i32>,  tensor<1x4xf32>) -> (tensor<1x2xf32>, tensor<1x2xf32>)
  %3:2 = "tf.Split"(%cst, %2#0) {device = "/CPU:0"} : (tensor<i32>,  tensor<1x2xf32>) -> (tensor<1x1xf32>, tensor<1x1xf32>)