Update CHARACTER_EMBEDDINGS.md

resources/docs/embeddings/CHARACTER_EMBEDDINGS.md

@@ -4,10 +4,7 @@
 are randomly initialized when you initialize the class, so they are not meaningful unless you train them on a specific
 downstream task.
 
-For instance, the standard sequence labeling architecture used by Lample et al. is a combination of classic word embeddings
-with task-trained character features. Normally this would require you to implement a
-[hierarchical embedding architecture](http://neuroner.com/NeuroNERengine_with_caption_no_figure.png) in which character-level
-embeddings for each word are computed using an RNN and then concatenated with word embeddings.
+For instance, the standard sequence labeling architecture used by [Lample et al. (2016)](https://www.aclweb.org/anthology/N16-1030) is a combination of classic word embeddings with task-trained character features. Normally this would require you to implement a [hierarchical embedding architecture](http://neuroner.com/NeuroNERengine_with_caption_no_figure.png) in which character-level embeddings for each word are computed using an RNN and then concatenated with word embeddings.
 
 In Flair, we simplify this by treating `CharacterEmbeddings` just like any other embedding class. To reproduce the
 Lample architecture, you need only combine them with standard `WordEmbeddings` in an embedding stack:
@@ -27,4 +24,4 @@ embedding = StackedEmbeddings(
 ```
 
 If you pass this stacked embedding to a train method, the character-level features will now automatically be trained
-for your downstream task.
+for your downstream task.