operand_range getArgOperands() {
      // Merge `in_tensors` and `captured_tensors`. `operandSegmentSizes`
      // has two elements, so it will be merging two operand ranges (the two
      // variadic arguments). Attributes like `batch_timeout_micros` will not
      // make it to the return array.

*   `tf.math`:

    *   Added `tf.math.approx_max_k` and `tf.math.approx_min_k` which are the
        optimized alternatives to `tf.math.top_k` on TPU. The performance
        difference ranges from 8 to 100 times depending on the size of k. When
        running on CPU and GPU, a non-optimized XLA kernel is used.

*   `tf.train`:

    *   Added `tf.train.TrackableView` which allows users to inspect the

		{"(*ComplexType).Size", Method, 0},
		{"(*ComplexType).String", Method, 0},
		{"(*Data).AddSection", Method, 14},
		{"(*Data).AddTypes", Method, 3},
		{"(*Data).LineReader", Method, 5},
		{"(*Data).Ranges", Method, 7},
		{"(*Data).Reader", Method, 0},
		{"(*Data).Type", Method, 0},
		{"(*DotDotDotType).Common", Method, 0},
		{"(*DotDotDotType).Size", Method, 0},
		{"(*DotDotDotType).String", Method, 0},