Chat:World/2020-06-28
Raghuveer: http://chat.codingame.com/pastebin/d7dd6358-0082-4de5-b9e2-6c364c611d97
Raghuveer: http://chat.codingame.com/pastebin/56651b52-a204-4e2f-8082-e45b4798fc7b
TheSpiffiest: Can anyone direct me to code for PID loops? I haven't done them before.
Raghuveer: i got the answer already
jacek: D:
Raghuveer: :what to do a clash:
tanmaycodernovice: https://www.codingame.com/clashofcode/clash/127308078ddb18c69a7e3c4632bf46daa7d7e06
N0Mana: Hi, how do I turn Java to C# in codingame puzzles?
LelouchVC2: There should be a drop down to change languages
LelouchVC2: Just click on where the language ur using is displayed
N0Mana: Oh there, ye I just didnt look hard enough. thanks!
LelouchVC2: <3
N0Mana: haven't used this account for years, and I've actually been learning C# in Unity.
codingame uses base C# right?
LelouchVC2: No idea
LelouchVC2: C# and Python run fast on here though, faster than C++
N0Mana: I see. Well looking at the frameworks, it does seem like base C#
struct: I dont know why you keep claiming that python runs faster than c++
Uljahn: ye, that's bs
LelouchVC2: Idk why Jesus had long hair
LelouchVC2: but some facts we have to acceptt
LelouchVC2: Btw, I never said Python runs faster than C++
LelouchVC2: I say specifically on *this* site
struct: Which is still not true
LelouchVC2: Which is clearly true
LelouchVC2: Go look for yourself
LelouchVC2: Start anything, and write the code in C++ and C#
LelouchVC2: Run through the test cases
LelouchVC2: C++ goes through them a few seconds per test
LelouchVC2: C# runs through them like butter
struct: compilation time?
Uljahn: lol
LelouchVC2: Why would it compile for every test?
struct: I dont know
LelouchVC2: I don't know why you attacked my claim if you didn't even test it yourself
struct: What is your definition of faster?
struct: For me compilation time doesnt matter
LelouchVC2: How could you possibly know if it compiles the code every time?
LelouchVC2: And even if it does, that means code is not optimized, which makes C++ still slower
LelouchVC2: "fast" is relative
LelouchVC2: On this site, fast would be how quickly ur code runs through the test cases
LelouchVC2: If this includes compilation time, so be it.
LelouchVC2: Real world, compilation time is once, and then the code will run nicely with C++. Run-time is whats important
LelouchVC2: But in this site, in CoC, when you've got a time limit and have to finish faster than others
struct: Well in multiplayers the compilation time is not taken into account
LelouchVC2: I wouldn't know, just pissed me off when I tried out C#
LelouchVC2: And it blasted through the test cases
LelouchVC2: I realize how disadvantaged I was in CoC
struct: Does the first test case take longer?
LelouchVC2: No, all equal time
LelouchVC2: if run-time on CG servers is not the issue
LelouchVC2: Then it's retarded that it's compiling it over and over again
Uljahn: CG engine is very old
LelouchVC2: omfg
LelouchVC2: No wonder I lost to ruby in fastest. we were both done in under 30 seconds
LelouchVC2: I just tested the solution with ruby, it literally just destroyed like 8 tests in 6 seconds
LelouchVC2: C++ took 6 seconds to finish like 2-3 tests
dbdr: I would be surprised if the compilation and running time mattered
dbdr: when you submit is what should matter
LelouchVC2: I think the point went over your head
LelouchVC2: If It takes me 3x longer to test my code
LelouchVC2: than someone else
dbdr: oh sure, that matters
LelouchVC2: :(
dbdr: use the best tool for the job, as always
MSmits: this has nothing to do with the engine though? C++ simply has to be recompiled when you change the code. Nothing weird about it. And yeah dev-speed suffers.
MSmits: kinda ballsy to do coc with c++ also :P
struct: MSmits but they did not need to recompile code for each test case
MSmits: thats true yes
MSmits: what happens when you test all at once?
MSmits: will it recompile each time?
MSmits: doubt that
struct: I guess
MSmits: I mean the test all button
MSmits: I can't imagine they would not use the same compiled code, would be silly to recompile
struct: well they do that in multis dont they?
MSmits: multis dont have a test all button
MSmits: its just a single test
jacek: can someone explain l2 norm. is it just adding lambda * sum(w^2) to the cost and nothing more?
jacek: im talking about code. fancy math mentions some derivative in weight update
jacek: but i think weight update should go as usual, the different cost function should cover it right?
jacek: or is it cost the same, but update is more like w[i] = w[i] + alpha*delta - 0.0001 * w[i]
jacek: doh i hate fancy math
dbdr: they don't recompile each time in multis anymore
MSmits: jacek are you throwing NN tensormath around now?
MSmits: what's this world coming to
Uljahn: do you mean l2 regularization?
jacek: eeyup
Uljahn: there are two ways i think, the second one is almost yours: https://visualstudiomagazine.com/articles/2017/09/01/neural-network-l2.aspx
Uljahn: so L1 decays weigths by a constant value while L2 decays by a percentage
jacek: thankies
Dick121380: When I was to participate in Clash of Code, it notified me that I must prove I'm not robot,but how to?:tired_face:
Dick121380: There is no new pages loaded
Uljahn: google captcha doesn't work in china
yshing: Get yourself out
MSmits: Dick121380 if you can't figure out, you must be a robot :P
WINWINWIN: :D
kdhakal: :grinning::grinning::grinning:
linjoehan: hey I'm puzzle of the week woo hoo
eulerscheZahl: congrats for becoming a puzzle linjoehan
MSmits: nice job, I'm just a test case
MSmits: or was it headcase? I forogt
MSmits: forgot
jacek: pacman puzzle of the week?
eulerscheZahl: checkers
jacek: pacman checkers?
eulerscheZahl: https://www.youtube.com/watch?v=Q2sDtTbzHiI rock paper scissors checkers - kind of...
linjoehan: what a great game
dbdr: the rules were not completely clear to me
Scarfield: if pawns have the same of RPS choose one, and fight agian. then there is one trap and a flag that apperently has to be captured
linjoehan: I think the trap can't move and the flag means you lose
linjoehan: or win
dbdr: sometimes you see the opponent kind
dbdr: but you can change it anyways
Scarfield: after a fight you know the winners kind, and if you fight them with the same "weapon" they can choose
Scarfield: otherwise they fight with the same as what the pawn won with last
dbdr: coming soon to CG?
Scarfield: too little FoW
Scarfield: you here Hjax? :)
Uljahn: that game looks similar to stratego
eulerscheZahl: "too little FoW"
- D
eulerscheZahl: https://www.codingame.com/contribute/view/507644dbe6bc6fc318eeaa25e26e48954355 oh, up for approval
eulerscheZahl: seems that he fixed the crashes on longer games
eulerscheZahl: 200 turns on the large map, 150kB replay size
eulerscheZahl: i'm sure this can be improved. But it's not crashing anymore
struct: yeah also noticed that
struct: if he rename sprites I think it will also change a lot
struct: ok it will only save 1846 chars
MSmits: is this 9x9 go? Or are the rules different?
MSmits: I see typo's in the statement also
eulerscheZahl: 13x13
MSmits: "Beside these simple rules there are a view more rules, on where stones can be placed and where not."
eulerscheZahl: a view
MSmits: I doubt there is any need to go beyond 9x9
tutubalin: this Rock-Paper-Scissors Stratego was an addon for ICQ
MSmits: it's nowhere near solved is it?
eulerscheZahl: yes, was an ICQ game
eulerscheZahl: i also remember a few other mini games. i really liked the pool game back then. and the war sheeps
jacek: 7x7 is weakly solved
tutubalin: Rock Paper Scissors https://www.youtube.com/watch?v=Jxbu80okI_U
jacek: storms \o/
dbdr: Atari go is supposed to be until the first capture
dbdr: not just smaller board compared to Go
jacek: atari go is smaller go fitted in atari 2600
eulerscheZahl: but how do i run it on my commodore?
jacek: play commodore go
dbdr: 2x2 go
darkhorse64: With handicap ?
brand: hello all
tutubalin: I prefer ZX Spectrum go
tutubalin: board is divided in 8x8 areas with 2 allowed colors per each
Yeet6969: bb
Yeet6969: no
Zenoscave: no 4004 go?
Ruukasu: Clash of code go go
Anthua: haha
WINWINWIN: Can someone explain how to save the output from a notebook on Kaggle?
AbundantPuddle: Did you try turning it off and turning it back on again?
WINWINWIN: :P
AbundantPuddle: I wish I actually could be more help, but I am not familiar with Kaggle at all.
WINWINWIN: Np :)