Building a classifier

I'm building a classifier in R to classify articles based on 5 categories(political articles, sports articles, etc)

I have a set of articles for which I know the classification of that I'm dividing into a training set and a validation set.

For each of these articles, I have a unique set of 512 decimal numbers called embeddings for each article that are supposed to aid me in building this classifier.

How would I go about building this classifier?

Would my predictors be the embeddings themselves? In other words, would they be the 512 decimal numbers I have for each article.

So my basic questions are when building this do I choose the features and how do I reduce dimensions.

What steps would I take when building this classifier that will be judged based on prediction accuracy?