TalkChess.com

Some more results with two nets from this thread against the latest sf dev. 1 thread, tc 1m+1s, the same 100 short normal openings for both matches. Bigger net is more solid.

This is my current net: Trained on depth 8 data. Validated with depth 19 data. https://drive.google.com/file/d/1P7EJuWPm02tKXx_aLxo_xwFWMSN3A44B/view?usp=sharing For use rename it to nn.bin and replace the original nn.bin in the "eval" folder. For use with: stockfish.nnue-learn-use-blas.halfkp_256x...

If anyone is interested, I analysed results pictured on fig. S4 in the Supplementary Materials ("Chess matches beginning from the 2016 TCEC world championship start positions"). It looks like each of 50 opening positions was played 100 times (50x with AZ White and 50x with Black). End result was +97...

Well, the search algorithm plays the move with the most visits. The initial visit is made to the move with the highest probability from the policy head. With a 1 visit "search", the search will evaluate the position after the move with the highest probability from the policy head, but no matter wha...

For me it's exciting but it's not that incredible as I was following Leela Zero in go and something similar happened there. I like playing against pure policy networks (1 playout, no search), it's a lot of fun in both chess and go. In go pure network is already around 3-4 dan level which would corr...

For me it's exciting but it's not that incredible as I was following Leela Zero in go and something similar happened there. I like playing against pure policy networks (1 playout, no search), it's a lot of fun in both chess and go. In go pure network is already around 3-4 dan level which would corr...

Remember that LCZ first began self-learning in February. It is now over 2900 CCRL. That is quite incredible if you ask me. I'm not attacking so I don't know why you go on defensive here. For me it's exciting but it's not that incredible as I was following Leela Zero in go and something similar happ...

I also tried a match of LCZ (the same HW and TC as above) against minimal SF9 (1 CPU, 32 MB hash, no book, no syzygy) with Noomen 2-moves openings and it was unfortunately almost absolute slaughter, around 96% for SF. Which Noomen 2-move openings? Can you give a link? This result doesn't seem surpr...

I can confirm that: - LCZ on GTX1060 and 1m+1s TC is now in CCRL 2900 ballpark (I tested ID228) when tested against engines in roughly 2900-3100 range, - cuDNN version is much faster but doesn't look any stronger than OpenCL one. I also tried a match of LCZ (the same HW and TC as above) against mini...

Website is nice but I think it would be really unique place to go if you could run gauntlets for LCZero. Its network is relatively small and should work really fine on CPU. If at some point it would be possible to add some decent GPU it could be really something.