-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Thanks for the very detailed report! SO good to see, that stronger
programs start using DCNN.
We should ask Nick, if he DCNN gets an exception from the KGS rules.
At the moment I would interpret them as not allowing multiple bots
using the same CNN, but of cause training this CNN is no magic and
only costs energy. For me it would be fine using it in tournaments!
Your factor: yes, I think this way it is nearly independent from the
factor (it just multiplies the final gamma with r, leaves the order
unchanged ...)
I use an aditive term gamma * (DCNN + z), but this was only a quick
shot too:)
Detlef
Am 18.09.2015 um 20:08 schrieb Hiroshi Yamashita:
> Hi,
>
> I tried Detlef's DCNN learning data with Aya.
> http://computer-go.org/pipermail/computer-go/2015-April/007573.html
>
>
I tested 1 playout/move selfplay, and DCNN with Aya got around
> 90% winrate. DCNN returns each move probabilty. I multiply it by
> 1000, and multiply it by each move's rating. (r *= 1000 means
> multiply by 1000).
>
> Test games are less than 100. But It seems muliply constant has no
> effect. 90% winrate is about +400 Elo. But this is selfplay and
> playout does not understand semeai(capture race). So I guess +50 or
> +100 Elo against human.
>
>
> 1 playout Aya with DCNN vs 1 playout Aya without DCNN. (1
> thread, selfplay, Xeon W3680 3.3GHz, GTS 450)
>
> winrate wins/games 0.943 83/88r *= 1000 0.897 78/87
> r *= 500 0.913 84/92r *= 200 0.932 82/88r *= 100
> 0.914 85/93r *= 50
>
> Select maximum uct_rave move. MM_gamma is each move's rating from
> Remi's Elo rating paper.
> --- r =
> result_DCNN(pos(x,y)); if ( r < 0.001 ) r = 0.001; r *= 1000;
> MM_gamma *= r;
>
> C = 0.31 ucb = moveWins/moveCount + C * sqrt( log(moveSum+1) /
> moveCount ); rave = raveWins/raveCount + C * sqrt(
> log((moveSum+1)*175) / ((moveSum+1)*0.48) );
>
> W1 = (1.0 / 0.9); // from fuego W2 = (1.0 / 2); beta =
> raveCount / (raveCount + moveCount * (W1 + W2 * raveCount));
>
> K = 1200; bias = 0.01 * log(1 + MM_gamma) * sqrt( K / (K +
> moveCount));
>
> ucb_rave = beta * rave + (1 - beta) * ucb + bias;
> ---
>
> Aya calls DCNN when node is created. Aya makes 900 nodes in 1
> playouts. GTS 450 needs 17.4ms for a position. 900*17.4 = 15.6 sec
> is needed. Aya needs 5 sec for 1 playout without DCNN, and 20.6
> sec with DCNN. So 4 times slower.
>
> I heard HiraBot jumped from 2d to 3d by using Detlef's data. He
> uses DCNN only in root node. HiraBot prediction rate without DCNN
> is 38.5%. MC_ark jumped from 2k to 1d by using Detlef's data.
> MC_ark uses DCNN only in root node and root's children. Aya's
> prediction rate is 38.8%, and Detlef's DCNN is 44%.
>
> Time for one position
>
> CUDA cores clock GTS 450 17.4 ms 192 783MHz GTX 970 1.6
> ms 1,6641050MHz
>
> *CPU235.0 ms ... Xeon W3680 3.3GHz one thread.
>
> GTX 970 is 11 times faster than GTS 450. Maybe it is equal CUDA
> cores ratio (8.6) x clock ratio(1.3). I also use caffe. Installing
> caffe was the most difficult part... And thank you Detlef for
> publishing your data!
>
> My test code and Makefile.
> http://yss-aya.com/20150907detlef_test.zip
>
> Regards, Hiroshi Yamashita
>
> ___ Computer-go mailing
> list Computer-go@computer-go.org
> http://computer-go.org/mailman/listinfo/computer-go
>
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.22 (GNU/Linux)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=K7kr
-END PGP SIGNATURE-
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go
