TalkChess.com

It's an MCTS node which is either from the GPU or the NNCache.

Hurnavich wrote: ↑

Fri Nov 09, 2018 8:25 am

Hi, do you have a link to the post of the most common 7 man suggestions

http://talkchess.com/forum3/viewtopic.p ... 17#p769925

Downloading all 7 men Syzygy is needless. Someone posted the most common 7 men endgames here a while ago, listing both the top in human and computer games. It's very top heavy. Most configurations just don't occur in practice. You can fit the most common ones in WDL format on a 500G or 1TB SSD. But ...

I don't disagree on the need to understand the inner workings but you will have a hard time beating vendor supplied optimized libraries such as Intel MKL, CuDNN, TensorRT etc... Lczero already tried the former approach first and eventualy switched to cuDNN and MKL blas etc. I am sure GCP put a lot ...

Another thing to notice in this table is that fp16 numbers tend to be >> 2x fp32 numbers.

This makes me a bit suspicious.

Do you support 7 men EGTB in this update?

Does the "smart TB access mode" work for them?

Updated information from Ankan on discord: with cudnn 7.3 and 411.63 driver available at nvidia.com minibatch-size=512, network id: 11250, go nodes 1000000 v0.17, default values for all other settings (2070 run was with v0.18.1 lc0 build but with same settings) fp32 fp16 GTX 1080Ti: 8996 - Titan V:...

Not seen any official announcement yet, but I imagine the answer is no. It was probably always a 'done deal' when it was accepted in the apparently secretive way it all happened at the last minute. Obviously a matter for TCEC and their reputation. leela-DX is an interesting project. But if I was th...

Also, your conclusion seems to confirm my feeling that Leela would suffer greatly in Chess960. Obviously, its opening-phase advantage over other engines would be nullified, but it's probably much more than that. If you train it to Chess960 too, it'd work fine. If I write a classical engine and make...

Looking at TCEC Div4 games, what is striking to me is not that NN eval is simply "too optimistic", as many have noted - it's that it often rises significantly as the game progresses, without a meaningful improvement of the position. This has more to do with the temperature used in training than any...