{"(*TextHandler).WithGroup", Method, 21},
		{"(Attr).Equal", Method, 21},
		{"(Attr).String", Method, 21},
		{"(Kind).String", Method, 21},
		{"(Level).Level", Method, 21},
		{"(Level).MarshalJSON", Method, 21},
		{"(Level).MarshalText", Method, 21},
		{"(Level).String", Method, 21},
		{"(Record).Attrs", Method, 21},
		{"(Record).Clone", Method, 21},
		{"(Record).NumAttrs", Method, 21},

node and associated metadata.}]>:$key,
    Res<TF_BoolTensor, [{If the `must_compile` attr is false the _XlaCompile op
can decide not to compile the clusters based on some profitability
heuristics.  In that case `compilation_successful` is false if _XlaCompile
chose not to compile the cluster.  If the `must_compile` attr is true then
_XlaCompile always attempts to compile the cluster and

            `Model.predict_step`.
        *   SavedModel uses its own `Model._saved_model_inputs_spec` attr now
            instead of relying on `Model.inputs` and `Model.input_names`, which
            are no longer set for subclass Models. This attr is set in eager,
            `tf.function`, and graph modes. This gets rid of the need for users