TalkChess.com

For my tensorflow script I use the numPy loading of a binary file with 1-byte per on-off input, it was surprisingly quick even for 10+ gigabyte files. I export the binary training data from my dev slow.exe which includes my auto-trainer code. My first try with tensorflow I loaded and converted text...

For training I use C++ with libTorch. I've been experimenting with PyTorch, the data-loader from PyTorch is so slow that it is unusable. Writing a data-loader in Python is not done either, I've never seen a language as slow as PyThon, It makes me wonder why most data-scientists insist on using this...

I've been working on a network with a similar structure as Jonathan is using. Currently it has 736 inputs, the layout is: 736:512x32x32x1. The results that I get with it are significantly better as with my HCE, maybe for a large part because my HCE is 10 years old and not up to modern standards. For...

Fun fact: in one of the games, my engines reached the KBN-K endgame, so this does actually happen. Since it doesn't support EGTBs, it's good that it has some KBN-K code to win this. Indeed sometimes this happens. I don't think that my engine is capable of winning this without TB. Actually I never t...

It was a remarkable tournament, probably one of the hardest monthly tournaments ever. I didn't expect Nightmare to win this tournament with LCzero and two RaspberryPI's tied for 2nd place. The difference in strength between the numbers 1 to 14 is probably negligible. Six weeks ago I've implemented a...

The monthly on-line engine blitz tourney for January 2021 will take place on: Saturday January 16, 2:00 PM EST (Boston time), 20:00 CET (Amsterdam time) To connect: winboard -zp -ics -icshost nightmare-chess.nl -icshelper timeseal -fcp ENGINE.exe -fd ENGINEFOLDER -autoKibitz (for UCI engines, add -f...

In every iteration I change the 300k pos by first shuffling the 1m+ training pos randomly. So you are using mini-batches for training. Is it a standard for Texel tuning or using a batch size equal to the size of the training set is also fine? (leaving time considerations aside) Using a batch size e...

To be honest I have the feeling that development tools under Linux are so cumbersome compared to Visual Studio that I don't want to step away from it. When I really like a piece of software I'm always willing to pay for it, I don't want everything for free. I have that feeling with visual studio: "...

The point is that it is all bull sh*t, with so many threads on slow CPU's the engines will only degrade.

In my case, with regard to development tools, I refuse to use anything that is paid or not cross platform. I don't want to write an engine that needs a huge code change to run on Linux or the Mac (even though I'll probably never have a Mac), and I don't want to change editors if I should ever switc...