Since they are given the playlist from the official Eurovision youtube channel, not all are judged by the studio version.
I do believe eurojury is a nice indicator, but it has its fails and sometimes just gives the wrong idea. In 2019 Michael Rice got a great eurojury score on his live performance, in the contest he got lousy 8 jury points.
Salvador Sobral was judged by his live performance. He was doing okay with eurojury, but 21/37 juries gave him 0 points, this never gave us indication he would landslide the real jury score. Now you can argue that his Eurovision performance was just better, but I would say pretty much every song gets eveluated from national final to eurovision.
Kristian Kostov was also underperforming for such a type of song. This gave the idea he would not do that great at Eurovision, yet he easily would have achieved more than 300 jury points if there was no Amar pelos dois.
There are tons of other examples.. like Eugent Pushpepa being rewarded by only one single eurojury member, but then getting a 7th place with juries in the contest. Or the United Kingdom being 5 times in the eurojury top 10 and not a single time managing to score a top 15 score at the contest. And the one time Eurojury didnt like UK (2017), they managed to get a top 10 result.
Not trying to dismiss eurojury. There are some nice findings. Austria 2018 got a 4th place, which was an indicator it could do really well (but then again, ZiBBZ only had 10 points less than Austria and didnt manage to qualify).
I still get the impression its actually more accurate if you try to put a potential jury ranking after all songs are out than relying on eurojury. Maybe I am completely wrong, but I have a hard time imagining Slovenia being higher than Lithuania in the juryscore of this semi (both judged by their live performance).
I also had Belgium easily in the top 10 jury of this semi.. going to be interesting how everything turns out in the end.