In addition to the other advice, I'd recommend starting with the "seminal" articles in topic of interest – articles that are heavily cited by other researchers

In addition to the other advice, I'd recommend starting with the "seminal" articles in topic of interest – articles that are heavily cited by other researchers – before moving on to something more fine-grained. Even though I've read a fair amount of academic literature, it'd be difficult for me to pull an article off of arXiv and understand what's going on, since much of academic writing is building off of prior work and requires context. Seminal works are usually written more clearly because the author is trying to introduce a new idea. Google scholar makes it a bit easier to identify seminal works, as the results return a count for "cited by". You'll see that some are cited thousands of times by other researchers. Plus, these are typically the most interesting articles to read either way.