Re: [Computer-go] DarkForest policy network training code is open-source now.
Hi Gian-Carlo Pascutto, My mistake. The model file didn't get pushed. It is now in train/rl_framework/examples/go/models/ Best, Yuandong On Wed, Oct 5, 2016 at 5:00 AM, <computer-go-requ...@computer-go.org> wrote: > Send Computer-go mailing list submissions to > computer-go@computer-go.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://computer-go.org/mailman/listinfo/computer-go > or, via email, send a message with subject or body 'help' to > computer-go-requ...@computer-go.org > > You can reach the person managing the list at > computer-go-ow...@computer-go.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Computer-go digest..." > > > Today's Topics: > >1. DarkForest policy network training code isopen-source now. > (Yuandong Tian) >2. Re: DarkForest policy network training code isopen-source > now. (Thomas Rohde) >3. Zen19K2 is strongest player on KGS (Hiroshi Yamashita) >4. Re: Zen19K2 is strongest player on KGS (uurtamo .) >5. Re: DarkForest policy network training code is open-source > now. (Gian-Carlo Pascutto) > > > -- > > Message: 1 > Date: Tue, 4 Oct 2016 14:47:30 -0700 > From: Yuandong Tian <yuandong.t...@gmail.com> > To: computer-go@computer-go.org > Subject: [Computer-go] DarkForest policy network training code is > open-source now. > Message-ID: > <CAA9V9Gxv98nDxTLk3FsK0aV+jy1JDJYg-YR47nUtxvuJd5MKUQ@ > mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Hi all, > > DarkForest training code is open source now. Hopefully it will help the > community. > > https://github.com/facebookresearch/darkforestGo > > With 4 GPUs, the training procedure gives 56.1% top-1 accuracy in KGS > dataset in 3.5 days, and 57.1% top-1 in 6.5 days (see the simple log > below). The parameters used are the following: --epoch_size 256000 --GPU 4 > --data_augmentation --alpha 0.1 --nthread 4 > > | Sun Aug 21 21:54:15 2016 | epoch 0001 | ms/batch 721 | train > [1pi@1]: 11.230860 [1pi@5]: 30.617970 [3pi@1]: 3.099219 [3pi@5]: > 14.042188 [2pi@5]: 18.935938 [2pi@1]: 4.482813 [policy]: 8.361849 > | test [1pi@1]: 27.767189 [1pi@5]: 59.403130 [3pi@1]: 5.380469 > [3pi@5]: 24.729689 [2pi@5]: 34.030472 [2pi@1]: 8.382812 [policy]: > 6.558414 | saved * > > | Thu Aug 25 10:35:11 2016 | epoch 0381 | ms/batch 719 | train > [1pi@1]: 56.226566 [1pi@5]: 87.523834 [3pi@1]: 21.542580 [3pi@5]: > 51.992970 [2pi@5]: 68.728127 [2pi@1]: 34.199612 [policy]: 3.736506 > | test [1pi@1]: 56.124222 [1pi@5]: 87.432816 [3pi@1]: 21.60 > [3pi@5]: 52.107815 [2pi@5]: 68.922661 [2pi@1]: 34.421875 [policy]: > 3.737540 | saved * > > | Sun Aug 28 00:49:32 2016 | epoch 0661 | ms/batch 721 | train > [1pi@1]: 57.075783 [1pi@5]: 88.215240 [3pi@1]: 22.512892 [3pi@5]: > 53.472267 [2pi@5]: 70.093361 [2pi@1]: 35.576565 [policy]: 3.638625 > | test [1pi@1]: 57.101566 [1pi@5]: 88.271095 [3pi@1]: 22.295313 > [3pi@5]: 53.226566 [2pi@5]: 70.085938 [2pi@1]: 35.185940 [policy]: > 3.646803 | saved > > Thanks! > > Best, > Yuandong > > > Yuandong Tian > Research Scientist, > Facebook Artificial Intelligence Research (FAIR) > Website: > https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ > -- next part -- > An HTML attachment was scrubbed... > URL: <http://computer-go.org/pipermail/computer-go/ > attachments/20161004/e96304c7/attachment-0001.html> > > -- > > Message: 2 > Date: Wed, 5 Oct 2016 01:56:18 +0200 > From: Thomas Rohde <t...@bonobo.com> > To: computer-go@computer-go.org > Subject: Re: [Computer-go] DarkForest policy network training code is > open-source now. > Message-ID: <461b50e8-bb6b-46bb-b481-074773308...@bonobo.com> > Content-Type: text/plain; charset=utf-8 > > Thank you, Yuangdong Tian, > > > On 2016-10-04 at 23:47, Yuandong Tian <yuandong.t...@gmail.com> wrote: > > > DarkForest training code is open source now. Hopefully it will help the > community. > > > > [..] > > spreading this wide and far, with high hopes that we’ll soon see many more > strong Go apps! > > > Greetings, Tom > -- > Thomas Rohde > Wiesenkamp 12, 29646 Bispingen > > t...@bonobo.com > > -- > > Message: 3 > Date: Wed, 5 Oct 2016 12:02:15 +0900 > From: "Hiroshi Yamashita" <y...@bd.mbn.or.jp> > To: <computer-go@computer-go
[Computer-go] DarkForest policy network training code is open-source now.
Hi all, DarkForest training code is open source now. Hopefully it will help the community. https://github.com/facebookresearch/darkforestGo With 4 GPUs, the training procedure gives 56.1% top-1 accuracy in KGS dataset in 3.5 days, and 57.1% top-1 in 6.5 days (see the simple log below). The parameters used are the following: --epoch_size 256000 --GPU 4 --data_augmentation --alpha 0.1 --nthread 4 | Sun Aug 21 21:54:15 2016 | epoch 0001 | ms/batch 721 | train [1pi@1]: 11.230860 [1pi@5]: 30.617970 [3pi@1]: 3.099219 [3pi@5]: 14.042188 [2pi@5]: 18.935938 [2pi@1]: 4.482813 [policy]: 8.361849 | test [1pi@1]: 27.767189 [1pi@5]: 59.403130 [3pi@1]: 5.380469 [3pi@5]: 24.729689 [2pi@5]: 34.030472 [2pi@1]: 8.382812 [policy]: 6.558414 | saved * | Thu Aug 25 10:35:11 2016 | epoch 0381 | ms/batch 719 | train [1pi@1]: 56.226566 [1pi@5]: 87.523834 [3pi@1]: 21.542580 [3pi@5]: 51.992970 [2pi@5]: 68.728127 [2pi@1]: 34.199612 [policy]: 3.736506 | test [1pi@1]: 56.124222 [1pi@5]: 87.432816 [3pi@1]: 21.60 [3pi@5]: 52.107815 [2pi@5]: 68.922661 [2pi@1]: 34.421875 [policy]: 3.737540 | saved * | Sun Aug 28 00:49:32 2016 | epoch 0661 | ms/batch 721 | train [1pi@1]: 57.075783 [1pi@5]: 88.215240 [3pi@1]: 22.512892 [3pi@5]: 53.472267 [2pi@5]: 70.093361 [2pi@1]: 35.576565 [policy]: 3.638625 | test [1pi@1]: 57.101566 [1pi@5]: 88.271095 [3pi@1]: 22.295313 [3pi@5]: 53.226566 [2pi@5]: 70.085938 [2pi@1]: 35.185940 [policy]: 3.646803 | saved Thanks! Best, Yuandong Yuandong Tian Research Scientist, Facebook Artificial Intelligence Research (FAIR) Website: https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
[Computer-go] DarkForest is open-source now.
Hi all, DarkForest Go engine is now public on the Github (pre-trained CNN models are also public). Hopefully it will help the community. https://github.com/facebookresearch/darkforestGo Thanks! Best, Yuandong Yuandong Tian Research Scientist, Facebook Artificial Intelligence Research (FAIR) Website: https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] Game Over
Congratulations to Aja & DeepMind team! Amazing results :) ---- Yuandong Tian Research Scientist, Facebook Artificial Intelligence Research (FAIR) Website: https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] Computer-go Digest, Vol 72, Issue 9
aking additional territory. > > Study a few simple examples of groups of strong stones with a few or > more opposing stones in the neighbourhood, and you notice that degrees > of connection and degrees of life can differ from each other. So > influence / thickness must be described at least by these two degrees. > Furthermore, the values differ for Black and White, so at least four > parameters are necessary for a complete description. > > You find my informal definitions here > http://senseis.xmp.net/?Influence > http://senseis.xmp.net/?Thickness > or more carefully in my books. For the precise parameters of connection > and life see > http://senseis.xmp.net/?NConnection > http://senseis.xmp.net/?NAlive > > Concepts of proximity should be called 'proximity' while concepts of > influence should be called 'influence'. Proximity maps / functions do > not explain influence except for the simplest examples in which all > stones are alive and the view is clear in every direction. > > Computer go can have various study purposes (such as training neural > nets or predicting the final colour control in a scoring position) and > some sort of function over all intersections assigning them a single > number may be convenient for fast numerical training, but do not forget > that such a simplication trains both correct and false information > without distinguishing them properly. If we want to become stronger > players or create stronger programs, we must distinguish correct from > false information. Therefore, replace 1-dimensional by multi-dimensional > values if the task is to assess current positions rather than final > scoring positions, in which one value is sufficient. > > -- > robert jasiek > > > -- > > Subject: Digest Footer > > ___ > Computer-go mailing list > Computer-go@computer-go.org > http://computer-go.org/mailman/listinfo/computer-go > > -- > > End of Computer-go Digest, Vol 72, Issue 9 > ** > -- Yuandong Tian Research Scientist, Facebook Artificial Intelligence Research (FAIR) Website: https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] Facebook Go AI.
Hi all, I am the first author of Facebook Go AI. Thanks for your interest! This is the first time I post a message here, so please forgive me if I mess up with anything. 1. The estimation of 1d-2d is based on the win rate of free game in the last 3 months (since darkforest launched in Aug). See Table 6 in the paper. For ranked game, its rank is definitely lower since people tend to play more seriously. It seems that now darkforest is 1k and darkfores1 is 1d. 2. Here is the Pachi 10k command line for no pondering. pachi -t =1 threads=8,pondering=0 For pondering, it is simply pachi -t =1 threads=8 In both cases, all the spatial patterns are properly loaded. See the following GTP response: W>> protocol_version Random seed: 1448000132 Loaded spatial dictionary of 1064482 patterns. Loaded 3021829 pattern-probability pairs. 3. We use pachi version 11.99 as shown in the following GTP response: W>> version W<< = 11.99 (Genjo-devel): If you believe you have won but I am still playing, please help me understand by capturing all dead stones. Anyone can send me 'winrate' in private chat to get my assessment of the position. Have a nice game! 4. Darkfores2 is still DCNN model and no search is involved. Thanks! If you have any comments, please let me know. ---- Yuandong Tian Research Scientist, Facebook Artificial Intelligence Research (FAIR) Website: https://research.facebook.com/researchers/1517678171821436/yuandong-tian/ ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go