Announcing GoReviewPartner - v0.15 with Pachi and PhoenixGo

[ez4u]

Any chance for an unbundled windows version? Those of us just updating already have the engines and may already be using Lela 11 as well.

[pnprog]

Hi!

Any chance for an unbundled windows version? Those of us just updating already have the engines and may already be using Lela 11 as well.

You have different choices:

• Download the GnuGo bundled, because it's the lightest, then get rid of GnuGo inside (delete the goreviewpartner-0.8_gnugo-3.8/gnugo-3.8 sub folder, clean the gnugo command parameter in the Settings) then configure it to use Leela instead
• Run from the python source
• I should made a Windows download without bot

Do you mean Leela 11 beta?

[pnprog]

Hi!

I made some improvement on the aesthetic of GRP, goban and stones.
Finally I did not used any image, but went for generated texture, the goban looks like this:

wood.png (39.15 KiB) Viewed 18684 times

Now, because this is "generated", it's easy to add a salt of randomness inside, so that every time, the goban will look different.
I did something similar for the stones, but it's less obvious:

stones.png (91.32 KiB) Viewed 18684 times

Notice that both gobans are different.

I still plane to play a bit with that (on my local version, the goban wood has "nodes", and the stones bright spot position change with the time...), but I am already quite satisfied.

[ez4u]

I like the work you are doing on this. However, should the 'grain' of your texture be oriented top to bottom rather than left to right?

[pnprog]

I like the work you are doing on this. However, should the 'grain' of your texture be oriented top to bottom rather than left to right?

Now that you mention it, I seem to only find representation of Goban with vertical "grain" (wood fiber), and parallel with the Goban grid.

Is there some sort of aesthetic convention for that?

By the way, I have been looking at the AQ bot today, I think I will be implementing support for that bot as well in the coming v0.9

[djhbrown]

aesthetic convention

the eye is naturally drawn to contrasts, which in the case of incidentals, can detract from understanding.

So yes, it would make seeing what's happening on the board easier by adopting the convention of vertical grain, and it would be helpful to make the grain very soft, so that it does not attract the eye. The same goes for marking the handicap points, gridlines and coordinate letters/numbers, which could be more muted.

The last move could be indicated by a small grey blob instead of a large bright triangle, which grabs attention.

The opposite applies to markups - the annotator wants to draw the viewer's attention to the letter "A", so it should stand out clearly; computers have given us colour - so letter marks could be temperature coloured... oh! you do use colours - blue for >50%win and red for <50% - that's great, but unfortunately bright red catches the eye more than dull blue; so how about bright green for >50% and dull red for <50% (with the idea that green is good (safe) and red is bad (dangerous))?

btw, all the various stats Leela provides are only of interest to Leela developer; all a player needs to know is win%

[jeromie]

I like the work you are doing on this. However, should the 'grain' of your texture be oriented top to bottom rather than left to right?

Now that you mention it, I seem to only find representation of Goban with vertical "grain" (wood fiber), and parallel with the Goban grid.

Is there some sort of aesthetic convention for that?

When sitting at a real go board, it is conventional to align the board so that the grain traces lines between you and your opponent. In Japanese style boards that are not perfectly square, this also maintains the correct orientation of the board.

At a computer, it is natural to think of yourself as playing from the bottom and your opponent from the top, so vertical grain maintains this convention.

[pnprog]

Thanks both for the explanation, I have change the wood fiber direction to be vertical now.

an option for batch processing several games (useful for processing a collection of games overnight).

I have just finished to implement that, but it had to be used in command line, with python.

Usage: python bot_analysis.py [--range=<range>] [--color=both] [--komi=<komi>] [--variation=<variation>] <sgf file1> <sgf file2> <sgf file3>

Here are a few examples:

Code: Select all

python leela_analysis.py mygame.sgf

Analysis of mygame.sgf, first variation, using Leela, all moves, both colors, komi as indicated by the SGF file

Code: Select all

python leela_analysis.py mygame1.sgf mygame2.sgf mygame3.sgf

Same as above, for the three SGF files will be analyzed

Code: Select all

python ray_analysis.py *.sgf

Same as above, but all the SGF files will be analyzed, and the analysis will be performed by Ray

Code: Select all

python ray_analysis.py --color=black mygame.sgf

Analysis of black moves only

Code: Select all

python aq_analysis.py --range="10-20, 44,46, 100-110" mygame.sgf

Analysis of moves 10 to 20, 44, 46 and 100 to 110 by AQ

Code: Select all

python aq_analysis.py --variation=3 --komi=7.5 mygame.sgf

Analysis of variation 3, with setting the komi at 7.5 (some SGF files do not include komi)

[pleiade67]

Awesome ! Thank you very much !

[pnprog]

Hi everybody!

Here comes GoReviewPartner version 0.9 !

The mains changes compared to v0.8 are:

• Added support for AQ, the new bot that apparently is stronger than Leela and Ray
• Lot of "eye candy" improvement (goban/stones, interface, logo)
• I put a lot of work into handling issues at run time (error when launching the bot, accessing/reading the config file or SGF file...) in order to intercept most of errors and fail gracefully (in a way that tell the user what happened)
• Added some scripting features for batch analysis (also useful for me to automatize testing)
• And a lot (really a lot) of code factorization, that will allow me to add support for new bots pretty easily in the future.

I made several download zip file for Windows (one for each bot), they can be found on the home page together with the updated documentation: http://ns1297.imingo.net/

For the next version, I am considering adding minimal support for other languages. I guess I can handle the French translation by myself, maybe also the Chinese translation. Then I would appreciate if some of you can support me for other languages.

[johnsmith]

Can we soon expect adding Leela Zero in the list of engines?

[roy7]

LZ isn't ready for use for reviewing, it's still very weak.

[yoyoma]

LZ isn't ready for use for reviewing, it's still very weak.

It's so that we can review LZ.

[johnsmith]

@roy7
I am aware that LZ isn't that strong yet, but it will be dan level before Christmas, don't you think? It wouldn't hurt to have it as a tool even now. GRP author mentioned that he had found a way to easily implement new engines. And GRP is my favourite tool in combination with Leela (Zero). I have a feeling that I could skyrocket my rank if I analyze games with GRP every day. Like a 9p sensei by my side at all times

@yoyoma
Always one step ahead

[djhbrown]

I have a feeling that I could skyrocket my rank if I analyze games with GRP every day.

i have a feeling you could skyrocket your rank if you asked Leela's advice before making a move; asking after is too late, because whereas children learn by imitation and Alfie learns by hill-climbing, the adult brain stubbornly refuses to believe that what it already believes could possibly be wrong even when faced with stark evidence of its flawed nature, so at best it can only learn cognitive dissonance. One evidence of this biological certainty is that you won't believe what i just said.