fix multi label doc

doc/advanced.md

@@ -14,8 +14,8 @@ This page will introduce some advanced usages in cxxnet, including:
  ```
  - In global setting, you need to specify how each field of the labels form a label vector. For example, we are interested in a localization task. In the task, we first need to output the label for one image, and next predict its position denoted by a bounding box. The configuration can be written as:
  ```
- label_vec[0, 1) = class
- label_vec[1, 5) = bounding_box
+ label_vec[0,1) = class
+ label_vec[1,5) = bounding_box
  ```
  - At last, in each loss layer, you need to specify the target of the loss:
  ```