Chat:World/2020-12-13

From CG community
Revision as of 11:57, 15 June 2021 by Chat Log (talk | contribs) (Created page with "<img src=/a/52739542369007> mzbear: that feeling when your 100+ line monstrosity of a function no longer has any TODO comments in it, so it should theoretically work now, but...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

mzbear: that feeling when your 100+ line monstrosity of a function no longer has any TODO comments in it, so it should theoretically work now, but you're too scared to test it

mzbear: oh, it doesn't even compile, but the editor no longer understand that because the error involves a lambda

Default avatar.png rwilson: that feeling when chat is dead and you could really use some commiseration

mzbear: that feeling when people use words you've never seen before

Default avatar.png rwilson: what does 'people' mean ?

mzbear: is this a philosophical question, or are you asking me if i'm referring to you? well, i had to look up "commiseration" from dictionary :)

Default avatar.png rwilson: ;) existential commentary .. basically ..

Default avatar.png rwilson: 'i feel ya bro '

mzbear: i'm afraid that particular sentence has a risk of being misinterpreted inappropriately due to ambiguity of "feeling" "ya bro"

Default avatar.png rwilson: ;) i feel that too ;)

Default avatar.png rwilson: ... in the totally not illegal interpretations ;) if taken literally

mzbear: lol

mzbear: something has gone horribly wrong with my code

mzbear: oh lol, i removed the code that read board state because i no longer needed it (i generated it from played moves instead) ... but forgot to leave the code that reads the inputs

Default avatar.png rwilson: ... is it wrong that I 'HEARD' the old mortal combat "WHOOPSIE" when i read that @mzbear ? ( i know it's 'toasty' but always sounded like whoopsie ;) i.e. mistake )

mzbear: lol

mzbear: now my bot is somehow playing invalid modes. great, just great

Default avatar.png rwilson: @mzbear which bot you working on ?

mzbear: Blocking

mzbear: i thought my move generation was working fine already .. ugh. what have i changed? oh, right, i rewrote roughly 300 lines without testing, i changed everything

mzbear: ah, valid move generation works fine, but my bot doesn't know how to play them on the board correctly

mzbear: yay, now it works! however, my bot plays incredibly badly, so i clearly have more bugs to fix :D

mzbear: argh this is performance issue ... monte carlo tree search isn't gonna do any good if i can only simulate 13 moves out of the 200 available. oh boy, optimization time

mzbear: maybe i should've gone for best-first minimax instead, to avoid having to do a full random play on every evaluation. my move generation is just too slow

YodaMaster123: Arent problems like this too basic?

YodaMaster123: v

YodaMaster123: https://www.codingame.com/contribute/view/589442e35a6b1ebca2f866fc05262591fb6e

MadKnight: naa we already got a x**3-3 one

YodaMaster123: That`s still not so bad...

YodaMaster123: print heads/tails is too ridiculous

MadKnight: good as a tutorial

MadKnight: pretty much the same as the tutorial puzzle

MadKnight: onboarding

YodaMaster123: Yeah, but at least the solution is not given in the problem statement.

MadKnight: suggest them to turn it into CoC

mzbear: it's for the "fastest" mode only, which basically means "how fast can you read the problem description?"

MadKnight: oh it is fastest already?

MadKnight: then good enough

MadKnight: hey mzbear

MadKnight: are u playing RAIC ?

mzbear: i dont know what RAIC is, so... no

YodaMaster123: Russian AI Cup?

MadKnight: it's another AI competition

MadKnight: join it

YodaMaster123: https://russianaicup.ru/

mzbear: can't afford distractions now, i wanna get my Blocking bot into wood 1 >_<

MadKnight: what game

mzbear: Blocking ... the puzzle of the week

MadKnight: oh

mzbear: it has a tad bit annoying move generation problem

MadKnight: u can join RAIC after u get wood1

YodaMaster123: Thanks @MadKnight_, I am joining the contest.

mzbear: i can look up RAIC afterwards, not promising anything

mzbear: i might find something more pleasant to do after this :D

MadKnight: nice

mzbear: compilers are so smart these days. half of the optimizations i'm trying produce slower code...

jacek: tryng to outsmart compiler eh

mzbear: it used to be much easier in the old days. this code isn't quite suitable for the type of optimizations the compilers today cant handle, either

Default avatar.png Maxim251: Thats true. I watch one week ago cppcon about compilers, and You will be suprised how smart compiler is now.

Default avatar.png Maxim251: If You do optimisation, for compiler will not do nothing, because without optimising code compiler is doing it for You anyway

mzbear: ffs i've lost 10% performance in recent changes and i cant remember what it was :D

Default avatar.png Maxim251: link to Youtube

Default avatar.png Maxim251: https://www.youtube.com/watch?v=w0sz5WbS5AM

Default avatar.png Maxim251: This is amasing showing how compiler works.

Default avatar.png Maxim251: in 12:34 he said also about performance

pascscha: petition to ban reverse mode from CoC

jacek: oO

eulerscheZahl: petition failed

pascscha: ok, was worth the try

jacek: he has no power here

eulerscheZahl: who?

jacek: how can you fail the petition? you have no authority here :v

jacek: tho personally i like reverse mode

eulerscheZahl: i don't make it fail, I just predict the outcome

pascscha: Reverse mode is just much more inconsistent, you either see it straight away or you don't. So I don't like it because it messes up my streaks that I get from fastest and shortest mode, but of course I fully understand that taking away revers is not an option. It just frustrates me immensely from time to time.

Default avatar.png Maxim251: i dint have idea how make this puzzle ANEO Sponsored Puzzle

jacek: a true reverse mode

Default avatar.png Maxim251: http://chat.codingame.com/pastebin/9bb05ff6-68ad-4a83-b929-480d50554c19

Default avatar.png Maxim251: http://chat.codingame.com/pastebin/fc371661-2c30-4d79-9cc3-c60678aef8e9

Default avatar.png Maxim251: in python is this one

Default avatar.png Maxim251: but what about 3 trafick lights?

Default avatar.png Maxim251: how to check it?

Default avatar.png Maxim251: here is explanation

Default avatar.png Maxim251: https://www.xarg.org/puzzle/codingame/aneo-sponsored-puzzle/

magurofly: Does anyone wants to participate today's ABC185? https://atcoder.jp/contests/abc185?lang=en

mzbear: argh, after refactoring, my bot is playing illegal move and running into infinite loops, and time is running out on the puzzle of the day...

Default avatar.png YusufEmad04: hello

twitlydoof: yo

jacek: puzzle of the week, you mean

mzbear: indeed, week not day. i cant even think straight lol, no wonder my code is buggy

Velcoro: thanks for the link, magurofly

mzbear: ffs. performance optimizations won't save me it seems, I need to implement RAVE into my MCTS

mzbear: i'm gonna run out of time :(

Stilgart: eulerscheZahl: i was thinking about removing them, just in case

Stilgart: but it won't be necessary

eulerscheZahl: continuing our #fr discussion here?

Stilgart: (might be a better place, indeeed)

eulerscheZahl: sorry, i'd rather write in english than let google butcher my sentence ;)

mzbear: https://youtu.be/KEkrWRHCDQU

tibithegreat: I was reading a bit on a forum thread about UTT and found this

tibithegreat: "you randomly get 2.2GHz, 2.4GHz or 3.0GHz CPU"

tibithegreat: Is this still true?

eulerscheZahl: not sure about the exact hardware but there are different machines in use

tibithegreat: that's a bit weird

tibithegreat: I would have assumed it was always the same hardware

mzbear: random hardware? oh boy, that would explain why my optimizations seemed erratic. but it's too late to set up a local development environment for this, i cant be arsed to write a referee in time

therealbeef: submissions are very erratic. yesterday i resubmitted a UTTT bot that was #5/Gold and it ended up #38. Same code.

therealbeef: It's quite frustrating to be honest

tibithegreat: yeah I encountered this as well, but I just assumed it boils down to the randomness in how opponents are chosen

tibithegreat: since it's not really transitive

tibithegreat: like A beats B beast C beats A

tibithegreat: and depending which opponents the system chooses for you, you can get quite different outcomes

therealbeef: true, but for uttt it plays 220 matches, and after 10% im usually playing the same top 40 as before until the end. rating dropping from 31 to 27, it's too much. i understand that some designs can handle speed drops better. if my bot loses 1 ply, it's quite dramatic, but those mcts things might not care too much

Westicles: level 55?

eulerscheZahl: yes, that was unnecessary :rolling_eyes:

Westicles: Hmm, well I think it is important to publish solutions. So few do it

mzbear: "those better than boss will be promoted" .... boss score: 30.70 ... the best player score, by author of the puzzle: 26.30

mzbear: .....

mzbear: i'm starting to think this game is flawed

mzbear: ohhhh ... super fun gotcha, the game end condition is not mentioned and i had assumed it wrong!!

mzbear: the game ends if a player with lower score has to skip turn, but NOT if player with higher score has to skip turn

mzbear: and of course, the provided "played moves" does not include a mention if some player has to skip their turn

eulerscheZahl: https://twitter.com/FDesmoulins/status/1336720001550716928 what is CG working on?

tibithegreat: looks like a pool table?

tibithegreat: or an arcade in general

tibithegreat: or airhockey actually

eulerscheZahl: looks like a more interactive game than what we know now

tibithegreat: maybe it'a a coop game vs a boss

tibithegreat: where 2 player's bots have to work together to beat a boss

tibithegreat: that would sound... fun :D

tibithegreat: it does seem like he is clicking in the game viewer

Westicles: probably mobile. kind of surprising this site isn't more phone friendly

mzbear: i'm upset that my 0.1ms handwritten heuristic is stronger than my 200ms MCTS

MuddySneakers: MCTS is difficult w/ the Blocking because the branching factor for the first few turns is humongous. You have to reduce the number of possible moves you consider.

mzbear: i was initially considering different algorithm, but since those "what you learned" checkboxes suggested MCTS i decided to go for it... grr

MuddySneakers: So, I'm using MCTS and I beat the boss. It's doable. But you can just use it vanilla out-of-the-box.

MuddySneakers: Sorry, *cannot*

mzbear: well, i just finished the third refactoring, now i can finally implement RAVE which should help with the early game

Default avatar.png Maxim251: Yes, I have solution!!

Default avatar.png Maxim251: ANEO Sponsored Puzzle

Default avatar.png Maxim251: Maybe is not my solution, because I couldnt understand it with modulo usage

Default avatar.png Maxim251: http://chat.codingame.com/pastebin/1e053072-891f-44c8-b5ca-3e38fabc4572

Default avatar.png Maxim251: And if 'if' statment is true, it means speed is to high. And then speed should be lowered, and whole operation begins again, thats why there is -1;

Default avatar.png Maxim251: I have lucky that somebody in ython make simple code, and I could rewrite it to c++.

mzbear: i really should have set up a local referee as soon as i got movegen working ... argh

mzbear: trying to develop this in the web editor only is too painful

struct: tibithegreat there are 5 machines

struct: two 2.2 Ghz, 1 of them has no lvl 3 cache, 1 2.4 Ghz with no lvl 3, 1 with 3.0Ghz and 1 with 3.1Ghz

struct: 2.2, 3.0, 3.1 all all have same lvl 3 cache

struct: The machine that performs the best for me is the 2.4Ghz one

eulerscheZahl: i attended a lecture called "high performance computing". the prof once told us he got a new PC and compared some program runtimes

LegendaryStone: reatoretch you mind sharing your code?

eulerscheZahl: usually it was about 1.2 times faster. except for one program that was more that twice as fast

eulerscheZahl: he then was wondering why that one is so much faster. studied the CPU manual

eulerscheZahl: and then he told us "i realized it was because of the LSD" :D

eulerscheZahl: he meant the loop stream detector

Default avatar.png reatoretch: ok. I just shared.

Astrobytes: lol euler

eulerscheZahl: a great prof btw

eulerscheZahl: once we had a broken computer and we left a note "don't use, defect"

eulerscheZahl: student writes below "challenge accepted"

Astrobytes: :D

eulerscheZahl: prof sees the response, takes his pen "I kill you (Texas chainsaw masarcre)"

Astrobytes: nutter

Astrobytes: I had highschool teachers like that, uni lecturers were all somewhat serious

eulerscheZahl: and he told us whenever we forget to write "return 0" in our main function, a little baby kitten gets run over by a lawn mower

Astrobytes: Yeah, someone said similar about citing sources

mzbear: ... submitted my bot to arena, found out it crashes if it isn't player 0 ....

Astrobytes: :)

mzbear: also found out that my bot crashes if any of its opponents crashes, because it refuses to skip turns if there are valid moves

jacek: big branching factor in the beginning? consider opening books...

mzbear: the pieces are chosen at random, too, making opening books a bit less useful

jacek: :c

mzbear: ahhh... due to branching, i cant fully simulate all players even for a single round when there's 3-4 of them, which messes up things.

mzbear: bugs fixed, time to try arena again

Default avatar.png Maxim251: for There is no Spoon - Episode 1 what better container use, vector of struct or 2D array of struct? Like NODE[x][y]?

struct: do you need to resize on there is no spoon?

mzbear: it has 30x30 maximum size, doesn't really matter how you do it

mzbear: there we go ... rank 7 in wood2 league .... so close and yet so far away

Marcaunon: when does the puzzlew of the week change?

eulerscheZahl: monday

Marcaunon: and what league to I need to get to count as "solved" for the home screen achievement?

Westicles: Wood1

Westicles: Otherwise you would already have it

Marcaunon: thnx

Gilad-Gur-Andelman: http://www.patience-is-a-virtue.org/

mzbear: dead end. after fourth refactoring, my code ends up crashing somewhere, and i cant find it .. and i dont have local referee to debug this locally

mzbear: the code is already 900 lines long, too. *sigh*

mzbear: normally i'd take a break and figure it out later, but .... when's the deadline?!

struct: game is avaialable after too

mzbear: but puzzle of the week!

struct: a tomorrow

mzbear: i'm investing in the sunk cost fallacy. i've spent so much time on this already that i dont want to fail narrowly :D

mzbear: waitamoment ... i'm using 4 bits for values 0-20 ... i'm an idiot

mzbear: well, would you look at that, it's no longer crashing

mzbear: still fails to beat boss, and the latest refactoring didn't improve performance either. fun

tibithegreat: that moment when you've been chasing a bug for a few hours, and you finally find it, and now you get to delete all the milion cerr's you added to find it

mzbear: cerr's did nothing to help find this. for some reason, the code still looked like it worked, most of the time

tibithegreat: I meant in my case :D

mzbear: and despite eliminating rand() ... it still went between 66k and 80k rounds in a loop before something went wrong, and i never managed to catch wtf happened

tibithegreat: I have to delete about 100 lines of debug :D

mzbear: ah, i see :)

mzbear: meanwhile, now i'm fiddlng random parameters around to see if i can make this work better ... except i dont have a testbench to do this, so this is pointless

mzbear: rank 3 in wood2 ... if i just wait it out, i might get lucky and win against the boss before the day is over. HMMM?

struct: boss is quite far away

struct: near 4.0 points

twitlydoof: Is codingame lagging?

mzbear: after some more tweaking, rank 1 in wood2 ... grrrrr

mzbear: gave up. went to chill in the livingroom, and the moment i relax i immediately got several new ideas for optimizations...

mzbear: i suppose i'll implement some of them

tibithegreat: I usually take my dogs for a walk :)

tibithegreat: when I feel blocked

tibithegreat: it's win-win cause I get to relax and see things from a different perspective, and my dogs get to go for a walk :D

Skynse: This year was a complete mess

Skynse: And I have a feeling that a portion of next year is going to be that way too

mzbear: haha none of my optimizations did any good. OHWELL

twitlydoof: every time i do shortmode

mzbear: that feeling when the optimization is too good to be true, and it is ... *sigh* ... 30% speedup between subsequent runs was due to hardware randomness

struct: yes its normal

struct: There are 5 machines

therealbeef: they only reliable way to optimize is to run your bot on your own computer against previous versions of the bot

therealbeef: or with profilers

mzbear: everytime i submit new revision, the boss loses a little bit of its score, even though my bot doesn't seem to be playing any stronger at all

mzbear: spent some time reading literature about the RAVE algorithm and then tweaked a few constants ... got much faster to high ranks, but still not doing too great against the boss :(

mzbear: found an interesting paper, too, an improvement on the RAVE algorithm called GRAVE ... but it involves one more tuning parameter, and i'm tuning these by hand in the web editor :D

struct: Whats RAVE?

struct: its for MCTS right?

mzbear: yeah. basically, in addition to updating MCTS win stastics, it updates all children along the backpropagation path if that move existed in the playout

mzbear: it helps deal with the ridiculous branching factor in this game

mzbear: because i get an additional statistic for many moves i didn't simulate

mzbear: boss score 29.52 ... me 29.28 ... GRRR

struct: its getting closer

mzbear: i'll just push this one constant a bit further and resubmit as-is lol

Astrobytes: Rapid Action Value Estimation struct. Never used it tbh

struct: I saw some paper that talked about it

struct: But I have trouble reading them

Astrobytes: Yeah, some of them are a bit... well, you know

Astrobytes: mzbear: do you need more sims?

Astrobytes: well, rollouts

mzbear: more sims would definitely help. but i'm already managing 1900 of them during the first round, so it's not that bad

mzbear: it gets much harsher for 4-player games, when the entire move tree gets obsolete before it's my turn again and it has to start from scratch

Astrobytes: I assume you've employed all the CG tricks?

Astrobytes: preallocating nodes etc

mzbear: they're preallocated yes

Astrobytes: cool

mzbear: the move generator is the bottleneck really

mzbear: i had to write two versions of it ... one that gives all moves, and one that gives an uniform random move

Astrobytes: did you profile it?

mzbear: nope, i'm still developing in the web editor only lol

Astrobytes: Stop that nonsense :D

mzbear: 915 lines of code currently ... in a 25x60 character window :D

Astrobytes: you know about CGLocal and CGSync right

mzbear: i saw something about them, and then it said about browser extension and i nope'd out for the time being

Astrobytes: well, fwiw they are useful

mzbear: i'm sure they are, this is quite painful way to develop really

Astrobytes: even if you're not running a local simulation just having your editor of choice on hand is a joyous experience

Astrobytes: Or copy-paste

mzbear: i'm finding that web editor quite handy, tbh

Astrobytes: Yeah, well, I find it OK for short things. And it's a heck of a lot better since they switched it to Monaco

Astrobytes: (ymmv depending on language, there are bugs and awfulness with the auto-suggest for a lot of users currently)

mzbear: some languages on CG are plain awful. I tried writing some puzzles in Pascal for some giggles, and for some reason the fpc is configured for 16 bit Integer type by default .. like MS-DOS mode

mzbear: and of course, no autocompletion or anything useful

Astrobytes: Glorious :D

Astrobytes: They're clearly going for the full retro experience.

mzbear: it's awful. I have a friend who writes fpc/delphi code primarily, and I was showing him codingame site two weeks ago and he asked about the pascal support...

Astrobytes: lol... well, you can put in a bug report/issue

Astrobytes: There are a few people who use Pascal on here sometimes iirc, T Bali is one

Astrobytes: So it's worth mentioning

mzbear: when does the puzzle of the week change? is it UTC time?

struct: its not at midnight

Astrobytes: I cannot remember if I'm honest, I rarely pay attention to it

struct: its at 13:00 CET I Think

struct: I could check email to be sure

mzbear: that's a lot of more time than i thought :O

ToshiTuringMachine: last 2 puzzles of the week were so good that you can play with them for years

ToshiTuringMachine: why rush next

mzbear: i want to reach wood1 while it's still the puzzle of the week :D

mzbear: no way i'd be putting this much effort into this otherwise

Astrobytes: ah, you should continue with more of the community-made games

struct: Just 3 games that are not worth playing

ToshiTuringMachine: there are also 96 pending contributions

Astrobytes: meh

Astrobytes: oh hey Toshi, how's it going

struct: Yavalath, Yinsh and Othello

Counterbalance: mzbear if you add {$mode objfpc} to the source it should alias integer to longint

mzbear: https://www.codingame.com/replay/515807626 ... omfg my bot is pathetic

ToshiTuringMachine: good Astrobytes and you?

Astrobytes: Cool. Yeah, all good here

ToshiTuringMachine: wow Counterbalance that's really a magic shot

ToshiTuringMachine: i was trying pascal too but stopped for a few reasons

Counterbalance: yeah it's like C++ #pragma ;)

ToshiTuringMachine: like... why de fak should one pick delphi/pascal in 2020

Counterbalance: you can probably enable delphi mode that way too

Counterbalance: ahyeah idk.. i stopped coding in pascal/delphi around 25 years ago

Counterbalance: it was my 2nd programming language so I do have fond memories

Astrobytes: wow, well I did not know that one tbh

ToshiTuringMachine: yeah me similar, they pushed it into high school as something better than basic

Counterbalance: basic being the first, indeed :)

Astrobytes: that was the standard progression back then :)

Counterbalance: you could do inline asm in pascal which was neat

ToshiTuringMachine: the very bad thing is that the compiler is not free

ToshiTuringMachine: commercial license

ToshiTuringMachine: when react flutter and all other are free... are they still living in year 2000?

Counterbalance: i had TP7, warez were all the rage since there were no internet laws to speak of as it was considered completely unreal and imaginary

Astrobytes: "warez" haha, I haven't heard that in a loooong time

Astrobytes: Toshi: capitalising on legacy codebase maintainers I guess

Counterbalance: yeah bbs time, pre internet :)

Astrobytes: I remember the BBSs, a friend of mine ran up the most incredible phone bill I have ever come across. His poor parents.

ToshiTuringMachine: how much?

Astrobytes: Off the top of my head I do not remember, it was in the thousands

Astrobytes: and they were very far from rich

Astrobytes: Also knew someone who did the same thing with 'adult' phone chatlines, but that was in the hundreds

Astrobytes: £ sterling btw

Zenoscave: did they change something with jabber rooms?

jacek: hm?

Astrobytes: rooms with jabbers in them

Zenoscave: I was added to a new room in the chat

Astrobytes: not sure tbh

Counterbalance: a bunch of numbered rooms?

Astrobytes: oh yeah, I forgot that's still happening

Counterbalance: I think they might be clash rooms from clash invites or something

Astrobytes: sometimes it's privatge message echoes

Astrobytes: *private

Astrobytes: Don't think they've quite figured it out yet. Possibly related to external clients iirc

Zenoscave: This one was named after my univserity

Astrobytes: Had you recently used it?

Zenoscave: never

Astrobytes: Or had anyone messaged you from it?

Zenoscave: nope

Zenoscave: it was empty

Astrobytes: There's been random numbered ones, duplicates of existing channels (just in web chat, not even counting jabber)

Zenoscave: weird

Zenoscave: Someone's doing funky things it seems

Astrobytes: Yeah. It's a 'known issue'

Zenoscave: but not funkadelic things it also seems

Astrobytes: There were some theories about that.

Astrobytes: hahah

Astrobytes: Ain't no P-Funk here

Zenoscave: nope

LastRick: Instead of asking "how long have you guys been coding?" just asked did you ever call a BBS? Talk about dating yourself

Astrobytes: Hahaha

mzbear: added small extra heuristic to MCTS to bias the initial move exploration towards larger pieces, so it more likely explores their moves first

mzbear: let's see if this is enough to push me to wood1....

mzbear: that seems to have been a good decision

mzbear: my score 29.28 ... boss 29.18 ... battles in progress 64%

LastRick: cutting it close :)

mzbear: oh no, it flipped back the other way around

mzbear: well, there's a chance i'll actually score higher than boss this time around

reCurse: Calling a BBS...

reCurse: Blast from the past

LastRick: yeah

LastRick: I swear, the dumber my algorithm on Mean Max, the better it does

mzbear: nnnooooo .... boss 29.01 .... me 28.92

LastRick: arrgggh,that stings a little

LastRick: what place

mzbear: i'm going to need an emergency can of pringles

VizGhar: what puzzle is it now?

mzbear: Blocking

mzbear: ... i'll just make that selection bias more aggressive and resubmit, lol

jacek: oO

mzbear: eh, looks like trying to force the search algorithm's hand is just making it weaker

Astrobytes: adjust exploration constant?

Astrobytes: *did you adjust

mzbear: i've tried adjusting it quite a bit, and sqrt(2) has worked the best so far ... going even 0.1 to either direction causes the bot to start playing noticeably weaker. then again, i haven't tried combining exploration tuning with the biased sampling yet...

jacek: do you use decisive/antidecisive move, mcts solver (though i dont know how that works in n-player games)

mzbear: i had some such code earlier, but didn't really manage to get it to work for me

Astrobytes: where's Smits when you need him eh

mzbear: this code has gone through massive amounts of bugfixing today, though

mzbear: i've been aiming for the low hanging fruit only because of the puzzle-of-the-week deadline

Astrobytes: yeah but now you've caught the Community Multiplayer Board Game Virus

struct: Othello boss is also strong

struct: Not made by me obviously

Astrobytes: that was darkhorse right? And he couldn't beat it initially iirc

DomiKo: is it that hard?

DomiKo: If I remember, pure MCTS can beat it

Astrobytes: Othello? Or Blocking?

DomiKo: Othello

DomiKo: I haven't done Blocking :(

Westicles: Quite the battle on CGfunge prime. I don't know how even to approach that

DomiKo: yea :/

Astrobytes: Hmm, not sure it was beatable with pure MCTS. I had an alphabeta

DomiKo: CGFunge isreally different problem

Astrobytes: still between dbd r and the other one at the top (forgot the name) ?

Westicles: xevi took back the lead

Astrobytes: ah

DomiKo: by a lot

Astrobytes: that is something

Astrobytes: I find the viewer to be awful on that (some colour-blindness), keep meaning to make PR

Astrobytes: *a PR

DomiKo: Westicles I can say that I will be approaching SameGame for the second time :D

Westicles: DomiKo great! There is still a bunch of room for improvement

DomiKo: I will try two algorithms

DomiKo: Wonder if I will get better results

darkhorse64: For Othello, i made a boss with a crippled engine but even with a faster one and a solver, it was to beat

Westicles: I try to run them every once in a while, but it isn't very satisfying. Let it run for 5 days and maybe it improves a little, maybe not

DomiKo: 5days is a lotttt

DomiKo: I only tried 12H for each testcase

DomiKo: I guess I have to buy some cloud to do it

Westicles: The guys on jsgames above me found those pretty quickly, so not necessary. Just need a good alg

Astrobytes: ah yes, I remember now darkhorse64

tibithegreat: when showing debug data to err, is the limit on the number of characters or on the number of lines? before it just prints "..."

tibithegreat: is there any way to bypass this limit?

DomiKo: ohhh another 16K in opti leaderboard

Astrobytes: DomiKo: you didn't like RAIC this year?

DomiKo: I didin't have time :(

DomiKo: hard time studying

Astrobytes: ah, fair enough. Best not to sacrifice the studying.

DomiKo: I had like 3 interesting contests is same time :(

Astrobytes: Oh there was a codeforces( or topcoder?) and aoc

DomiKo: Polish contest

DomiKo: https://potyczki.mimuw.edu.pl/

Westicles: azspcs got cancelled, so now I need something else to do

Astrobytes: wow, why did that get cancelled?

Westicles: He said he's taking some time off

Astrobytes: oh nice DomiKo

Astrobytes: Hmm

Astrobytes: Guess 2020 has been unkind to almost all of us

DomiKo: So I'm really happy that break is coiming

DomiKo: coming*

Astrobytes: Time to relax a bit eh

DomiKo: yeah, I need to recharge my batteries

Astrobytes: It's important. That brain needs a rest now and then you know

VizGhar: CG is like you have crush on somebody :D you simply won't let go

Westicles: They say programming is crack for the highly intelligent

VizGhar: Yeah... and imagine being addicted on your family and running

LastRick: Love running actually. Amazing how often I get an idea for CG when I'm on a long run

VizGhar: :thumbsup:

Astrobytes: I used to run a lot but due to the ongoing back issue I cannot sadly

Astrobytes: I find a musical interlude to be useful (playing or listening)

Astrobytes: Anyway, I'm done for tonight. SEe you all later

DomiKo: See ya

LastRick: later

LastRick: And here we are again. The ol' spinning wheel time of night.

Jasperr: :D

Default avatar.png Red_Hood_Jason_Todd: help I am 10 years old and I think the introduction went too fast

Default avatar.png Red_Hood_Jason_Todd: please somebody

LastRick: If you have a specific question, put it in the forum and someone can answer it

Lobster_Speed: sup

Lobster_Speed: no i dont think