// scratch1 = rsqrt(var + epsilon)
      RankedTensorType scalar_float =
          tensorflow::GetTypeFromTFTensorShape({}, kernel_type);
      auto epsilon = rewriter.create<ConstantOp>(
          loc, DenseFPElementsAttr::get(scalar_float, {op.getEpsilon()}));
      auto add_op = rewriter.create<chlo::BroadcastAddOp>(
          loc, var, epsilon.getResult(), scalar_broadcast_dims);