The final ranking in Leaderboards will be the position when the time finish. Or they will keep running for hours to set the final leaderboards ?
Here I was thinking this was a topic about how you can have 200 people ranked worse than you if your code doesn’t properly compile… >.>’
Do you have an example of a bot that doesn’t compile and is not in the last 200 bots?
The algorithm we use computes a score as a float, so it’s hard to have draws even if theoretically people that can’t move (timeout/invalid action/don’t compile) should have the same score. In practice, currently, the are 80 bots with a score between 3.45 and 3.60 that either doesn’t compile or fail at the first move.
You won’t be able to submit new code but we will wait until every AI is ranked.
No current examples.
Sometimes I’ll try to work and mess with how my bot is working for this current contest… but run out of time I can use on my current method of internet. And worried about losing my changes when I leave the IDE screen, I opt to submit it to the arena–and tell myself I’ll return to it the next day.
The results I return to always have me wondering why I am doing so well… I mean… for what I’ve submitted as a “just fuck it, I’ll make it not glitch out later” to actually place ahead of so many other entries…
Based on my experience, you don’t have to submit to arena to “save”.
Just running a few tests saves it there.
Then you have two copies - one in arena and one in IDE.
When you come back you have the IDE copy saved. but not the arena one. You may have to save locally for that.
You can have fights between them as well. (That’s when I found this.)