SPCC: Testrun of Lc0 Net 32930 finished

[pohl4711]

Testrun of Lc0 0.20.2 N:32930 finished.
Next testrun: LeelaFish v0.3 with the same Net (32930) and Houdini 6 as Aux-Engine (with Threads=1, Depth = 14 and Threshold = 20), because an early intermediate result of LeelaFish in my long thinking-time tournament looks very promising... take a look on the long thinking-time tournament !!!

https://www.sp-cc.de


(Perhaps you have to clear your browsercache or reload the website)

[Jouni]

Why not Houdini with tactical mode - maybe even stronger combo? But hybrids are not suitable for any tournaments and dubious even for rating lists.

[Leo]

Stephan is the only person to include Brainfish in his testing's. I like that. Even though you couldnt include the hybrid in a tournament we might as well see what it can do.

[Leo]

Why not Houdini with tactical mode - maybe even stronger combo? But hybrids are not suitable for any tournaments and dubious even for rating lists.

Good idea that I haven't heard before. Combine lcz with Houdini Tactical. Robert Houdarts clame to fame after all these years.

[pohl4711]

Why not Houdini with tactical mode - maybe even stronger combo? But hybrids are not suitable for any tournaments and dubious even for rating lists.

Because I made some pre-tests, in which Houdini tactical gave weaker results in LeelaFish, than Houdini default.

[pohl4711]

Stephan is the only person to include Brainfish in his testing's. I like that.

Thanx. As far as I know, Thomas Zipproth plans to release a new Cerebellum-version soon. Then I will do a testrun with latest Brainfish and Cerebellum, of course.

[Leo]

Why not Houdini with tactical mode - maybe even stronger combo? But hybrids are not suitable for any tournaments and dubious even for rating lists.

Because I made some pre-tests, in which Houdini tactical gave weaker results in LeelaFish, than Houdini default.

Interesting.

[Leo]

Stephan is the only person to include Brainfish in his testing's. I like that.

Thanx. As far as I know, Thomas Zipproth plans to release a new Cerebellum-version soon. Then I will do a testrun with latest Brainfish and Cerebellum, of course.

Looking forward to that.

[jjoshua2]

Testrun of Lc0 0.20.2 N:32930 finished.

I was wondering if you used the defaults or the Superfinal settings that improved tactical performance? Cpuct=3.4 and cpuctbase=10000 (and enough nncache for longer thinks).
(I've tested Leelafish some and it seems that it's actually worse with these tactical settings.)