View source for Talk:Firestarter
|Thread title||Replies||Last modified|
|Will we have a new king?||9||20:10, 2 October 2017|
|Roborumble participants||1||09:33, 7 August 2017|
|One on One Scores Down?||2||17:19, 6 August 2017|
|1.3||4||18:51, 1 August 2017|
|Wow!||5||17:15, 17 July 2017|
Maybe after those last pairings and a few more battles (to make sure) we can yell "Long live the King!" again? :P
Congrats on the last improvements, Firestarter is now turning into a monster both in 1v1 and melee!
You do not have permission to edit this page, for the following reasons:
- The action you have requested is limited to users in the group: Users.
- You must confirm your email address before editing pages. Please set and validate your email address through your user preferences.
You can view and copy the source of this page.Rsalesc (talk)
Thanks! Yeah, the melee difference between Firestarter and Neuromancer is small now. My next task will be to improve the 1vs1 score against top bots. I saw Knight beats a lot of strong bots quite easily already, congratulations on that :)
just noticed that Firestarter is just 0.06 APS behind Neuromancer! Congratulations on that improvement after your hard work! And it's also escalating very fast in 1v1! Great work, man!
Why didn't you add 1.4 to roborumble, any problems with it?
One on One scores are worser than the previous version. Do you know why is it like that?
I optimized the energy management for melee, so maybe Firestarter is too survivalist in One on One, not inflicting enough damage to score well, and I might have to create a different energy management for One on One. Not sure though, there might be other reasons.
Great result with 1.3! I might have to dust off my Neuromancer code again ;-)
Any idea what caused the improvements? Were you skipping a lot of turns before?
Thanks, we'll see, Neuromancer is still ahead by a big margin :P
Yeah, the gun was slow and it usually skipped a few turns per round, but I didn't expect it to make such a big difference. I also added a dynamic energy management dependent on estimated hit probability.
I really like just hearing that that's even conceivable to you still. =) It actually crossed my mind pretty recently, but I don't think it's likely to happen for me any time soon.
I have all my code setup to work with a simple build script, so it is fairly easy to get it up again. Although I've gotten lazy with IDEs that do indexing, which I never used before, so maybe I should try to get that working first. Neuromancer needs a bit of cleanup, and its multi-wave surfing in non-melee situations is very... CassiusClay shall we say =). No branching, which is something I'd like to change, but somehow still not skip turns.
Just wanted to say congrats on a really amazing result! At the moment, you are in 4th, but really not far at all from 2nd place. Excited to see where it settles.
I agree, this is a very impressive result. My Melee chops are seriously lacking, so it's even more impressive to me.
Thank you guys! Yeah, I am very happy about the result. Hopefully it can be improved some more, since I still have a lot of tweaking ideas.
Congratulations! Looks like you have a record of the fastest arrival to the top. Looking forward to see the improved version of the bot.