Closed HackYardo closed 3 years ago
Thank you very much for this offer which I happily accept.
I attached the translations file, which includes one block for each english text. I added a chinese translation (language code: 'zh') from google in the first block as an example of the required format.
The translations are given in unicode notation. To convert chinese glyphs into unicode you can use this website https://www.branah.com/unicode-converter.
Some texts have to be broken into multiple lines because they are too long for the available space. This is indicated by line-breaks \n. This will be necessary for some chinese translations, too. You can get an indication of how much fits into one line by looking at the positions of the \n in the english, korean and japanese texts.
translationsAddZH.txt BeCareful : the file name changed
Sorry for an error, in 'visits/move in analysis', the 2nd char is not 65f6, it should be 6790
This translation needs considering WeiQi/Go/Baduk, GTP, English&Chinese, AI(especially MCTS), KataGo, basic programming, and final user-friendly expression.
Some words and how they be translated: move(n) : "手" "着" move(v) : "行棋" "落子" input : "落子", the same to move(v) play(n) : "对弈" play(v) : "弈" "执" position : "盘面" "位置" own : "形势", for locution pass : "虚手", for locution undo move : "悔棋" ponder : "分析", the same to 'analyze' evaluate : "计算", the same to 'compute' or 'calculate', just for simplification analysisWideRootNoise: "分析深广度", which means the depth and width of analyzing playoutDoublingAdvantage : "保守激进度", which means being conservative or radical playout : "模拟对局" "新增模拟对局" visit : "总模拟对局", the difference between playout and visit is: Playouts and visits are both counting the same thing (one more leaf node in the tree of explored variations) but playouts are a delta per move, whilst visits are the total across tree reuse from previous moves.
From: https://lifein19x19.com/viewtopic.php?f=18&t=17195&start=100
https://github.com/lightvector/KataGo/blob/master/docs/Analysis_Engine.md
http://www.lysator.liu.se/~gunnar/gtp/gtp2-spec-draft2/gtp2-spec.html#SECTION000311000000000000000
Thank you very much for putting so much thought into the translations. I will include them in the next release.
I debug the translation file again through https://www.branah.com/unicode-converter Hope my work doesn't clash your work. Its version is the final version before next BadukAI release. translationsAddZHver2.txt
I translate the index.html too, here it is: indexZH.html.txt
By the way, what does 'n/s' mean? 'nodes per seconds'? Is it the same to po/s or v/s? Or something else?
Thank you very much for translating index.html, too, that's really helpful.
"n/s" indeed means "nodes per second". I see this often when speed is measured. I never found an exact definition, but I assume it's the same as v/s or po/s.
Because it seems that 'winrate' and 'histogram' are on the same level, I suggest you change the places:
histogram: ▢ △-score, upto ± ▭ ▢ winrate ▢ △-winrate, upto ± ▭ |
---|
\/
Histogram:
\t ▢ winrate
\t ▢ △-winrate, upto ± ▭
\t ▢ △-score, upto ± ▭
And you can put some other buttons in 'Edit config'.
What is the meaning of the parameter 'numMovesToRandomize' in LeelaZero? I can't find it out by Bing or Github search engine...
Why LeelaZero's analyzing (via 'ponder' & 'analyse all') is capped by 'maxPlayout' & 'maxVisit' while KataGo is not? I set maxPo = 1 and maxV = 1, KataGo's analyzing not changes only when 'v/m in analysis' change.
How about saving sgf with automatically adding .sgf suffix?
Can't read sgf files created by other Go APPs, I don't know why. For example: (;FF[4]CA[gb2321]SZ[19]id[2020-12-02]AP[wqbd]GN[#2020]DT[2020]PC[]AN[]KM[7]HA[0]PB[L]PW[T]BR[]WR[]RE[]N[@围棋宝典 ]C[#2020 黑 L VS 白 T ];B[pd];W[pp];B[dd];W[dp];B[cn];W[fq];B[qq];W[pq];B[qp];B[bd];W[lo];B[ee];W[ff];B[em];W[mr])
At last, Could you add 'translations.txt' to BadukAI repo? So that I could follow the last version of BadukAI, update chinese translation and avoid clashes between us. Moreover, I find tow translation errors in index_zh.html, and this html needs adding string "English Version" in the top-right corner!
Some new items of 1.8.0 can't display integrally on the little size screen of my phone...
Your suggestion certainly looks much nicer, but it takes four lines instead of two. And the settings menu is already as long as the screen on phones with 16:9 aspect ratio (I checked with my old one). That's why I squeezed it into two lines (and as all go players know, squeezing can make bad shape). Joking aside, I can make the settings menu a little wider (even on your screenshot there is still space to the left), then the text should fit.
What is the meaning of the parameter 'numMovesToRandomize' in LeelaZero?
It controls the parameter -m (meaning: "Play more randomly the first x moves"). That one doesn't have a telling short name, so I had to invent one (that's why you can't find it on search engines)
Why LeelaZero's analyzing (via 'ponder' & 'analyse all') is capped by 'maxPlayout' & 'maxVisit' while KataGo is not?
KG and LZ behave differently here. BadukAI puts "maxVisits" and "maxPlayouts" in the config in both cases. For LZ this restricts both play and analysis. For KG this restricts only play.
How about saving sgf with automatically adding .sgf suffix?
Good idea.
Can't read sgf files created by other Go APPs, I don't know why. For example: ...
There are two problems in the sgf file: 1) the "id" property: The parser doesn't like it (and I could not find it in the sgf specification, maybe some kind of extension ?) 2) the "CA" property: The parser does not know the encoding "gb2321" (probably it is limited to western text encodings)
If I delete the parts "CA[gb2321]" and "id[2020-12-02]" with a text editor, BadukAI loads the file correctly.
At last, Could you add 'translations.txt' to BadukAI repo? ...
I will do so. As far as index_zh.html is concerned, it is also missing the new part where I explain the display of differences in the histogram (since I added this after you translated).
index_zh.html.txt I update the index_zh.html from 1.7.0 to 1.8.0, you can just use it instead of the old one.
I uploaded the new index_zh.html
I find a big error in index_zh.html, the two screenshots of BadukAI should be zh rather than en!
Can I add zh extension after the end of index_zh.html to introduce something out of index.html? For example, kata1-...-680... in release 1.8.1.
Can BadukAI ignore "id" and "CA" for reading sgf file created by some other Go APPs?
If the buttons of settings are wider and wider, which are enough to contain explanation, shall we add them? For example:
kyu-rank |
---|
\/ kyu-rank(30=30K, -8=9D)
I find a big error in index_zh.html, the two screenshots of BadukAI should be zh rather than en!
I will make chinese screenshots when I do the next tests with chinese version.
Can I add zh extension after the end of index_zh.html to introduce something out of index.html? For example, kata1-...-680... in release 1.8.1.
English and chinese version should have same content, but it's a good idea to add a hint where to find additional models in both of them.
Can BadukAI ignore "id" and "CA" for reading sgf file created by some other Go APPs?
I didn't write the sgf parser myself, it's a standard package. I will have a deeper look into it and see what I can do.
If the buttons of settings are wider and wider, which are enough to contain explanation, shall we add them? For example: ...
Yes, good idea, but in the case of kyu-rank I would prefer something like kyu-rank (30=30K, -1=2D etc) since especially the high dan ranks are very dubious, so I wouldn't like to mention them explicitly.
If you are not worried, I make index_zh.html line to line index.html. It maybe helpful in compare they two. And it will be used as a supplementary of git diff. index_zh.html.line2line.txt
BTW, how about use LZ 287(the last one) rather than LZ 286?
I checked in the new index_zh.html.
As for LZ 287, I might do it some day, but I don't consider this urgent, as the Elo difference to LZ 286 is very small.
Thanks. The next index_zh.html to follow index.html, I will request a pull to BadukAI repo.
Because BadukAI has not many items, I could translate into Chinese in a day.