Fix typo in vector search section: #the-representation-revolution

qdrant-landing/content/documentation/overview/vector-search.md

@@ -46,7 +46,7 @@ The latest research in Machine Learning for NLP is heavily focused on training D
 
 We can take those **dense vectors** produced by the network and use them as a **different data representation**. They are dense because neural networks will rarely produce zeros at any position. In contrary to sparse ones, they have a relatively low dimensionality — hundreds or a few thousand only. Unfortunately, if we want to have a look and understand the content of the document by looking at the vector it’s no longer possible. Dimensions are no longer representing the presence of specific words.
 
-Dense vectors can capture the meaning, not the words used in a text. That being said, **Large Language Models can automatically handle synonyms**. Moreso, since those neural networks might have been trained with multilingual corpora, they translate the same sentence, written in different languages, to similar vector representations, also called **embeddings**. And we can compare them to find similar pieces of text by calculating the distance to other vectors in our database.
+Dense vectors can capture the meaning, not the words used in a text. That being said, **Large Language Models can automatically handle synonyms**. More so, since those neural networks might have been trained with multilingual corpora, they translate the same sentence, written in different languages, to similar vector representations, also called **embeddings**. And we can compare them to find similar pieces of text by calculating the distance to other vectors in our database.
 
 {{< figure src=/docs/gettingstarted/input.png caption="Input queries contain different words, but they are still converted into similar vector representations, because the neural encoder can capture the meaning of the sentences. That feature can capture synonyms but also different languages.." >}}