Search found 3129 matches

by Daniel Shawul
Sun Oct 14, 2018 9:54 pm
Forum: Computer Chess Club: General Topics
Topic: Scorpio 2.8.9 with TensorRT support
Replies: 4
Views: 796

Re: Scorpio 2.8.9 with TensorRT support

The last working Scorpio Edition on my computer Intel i5 Quad , 4 GB RAM (4 CPUs) is v2.8.6 for Windows 7 x64 (both executable files of each version) 2.8.7-2.8.8-2.8.9 can be embedded under WinBoard Interface or Arena GUI, but before the first ply the engine crash each time: unhandled exeception er...
by Daniel Shawul
Sun Oct 14, 2018 9:51 pm
Forum: Computer Chess Club: General Topics
Topic: Scorpio 2.8.9 with TensorRT support
Replies: 4
Views: 796

Re: Scorpio 2.8.9 with TensorRT support

The last working Scorpio Edition on my computer Intel i5 Quad , 4 GB RAM (4 CPUs) is v2.8.6 for Windows 7 x64 (both executable files of each version) 2.8.7-2.8.8-2.8.9 can be embedded under WinBoard Interface or Arena GUI, but before the first ply the engine crash each time: unhandled exeception er...
by Daniel Shawul
Fri Oct 12, 2018 9:19 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 23
Views: 4212

Re: 3 million games for training neural networks

Are those 2M positions available? Those are 770k position from quiet_labeled.epd (zurichess), and 1.x million pos from Alvaro that everybody here has. I don't know why I am not able to beat those networks from PGN-trained networks so far... I just tried Cheng's 12 million positions metnioned in thi...
by Daniel Shawul
Fri Oct 12, 2018 3:41 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 23
Views: 4212

Re: 3 million games for training neural networks

Perhaps collections like high level TCEC games would be useful for training. They have scores attached to each position. That should give very high quality answers for the actual value of a position. Yes, I think that is what I need. Training from PGN games using the result tag just doesn't seem to...
by Daniel Shawul
Thu Oct 11, 2018 10:04 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 23
Views: 4212

Re: 3 million games for training neural networks

Hi, I had my 8-core Ryzen 7 computer spend about 2 months generating quick Stockfish-vs-Stockfish games. The result is 3 million games that I think could be used for training neural networks, or to tune evaluation functions. I have the games available in PGN format with comments that include the sc...
by Daniel Shawul
Thu Oct 11, 2018 6:20 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hacking with Winograd
Replies: 10
Views: 700

Re: Hacking with Winograd

Henk is in love with his neurons :) Why reinvent the wheel when there are ton of deep learning libraries to choose from. Because these 'open source' libraries are never optimal for the things you want to do with it. Most people are lazy, basically writing a script in Python to use with Tensorflow/K...
by Daniel Shawul
Thu Oct 11, 2018 5:47 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Texel tuning speed
Replies: 47
Views: 3565

Re: Texel tuning speed

I am trying Texel tuning and have a question about expected speed of the method. I seems Weini is able to run the qsearch needed for each position and each evaluation of the error in around 0.06 millisecond. Let's say I have only 100 000 positions and want to optimize 10 parameters. It will requier...
by Daniel Shawul
Thu Oct 11, 2018 5:29 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hacking with Winograd
Replies: 10
Views: 700

Re: Hacking with Winograd

Henk is in love with his neurons :)

Why reinvent the wheel when there are ton of deep learning libraries to choose from.
by Daniel Shawul
Thu Oct 11, 2018 5:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Minimal network
Replies: 10
Views: 776

Re: Minimal network

So far I found that to learn only 4000 training examples I already need about 29 layers with 4 filters per layer in an ordinary convolutional neural network. To fit more examples I expect more layers and filters. Probably I need to switch to residual network too when number of layers increases. Ove...
by Daniel Shawul
Thu Oct 11, 2018 5:04 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Texel tuning speed
Replies: 47
Views: 3565

Re: Texel tuning speed

Thanks Jon, I have got them now.

I started filtering the positions for quietness but it seems most are quiet so I am going to use them they are.

Daniel