OpenChess

OrgZ wrote: Sat Nov 18, 2023 4:28 pm

Homayoun wrote: Thu Nov 16, 2023 1:31 pm I thought this engine uses two books by default now. I made a mistake. Anyway, Raid plays very well.
Greetings

Nah, u just load them bin books as Blood.bin and sport.bin been away for a bit Mr H. working on updates, it's a process, trying to make patzer AI less positional and more fun, hv archived that but it alters his strength in long TC.....so back to the drawing board. As for Raid his ready, but will release after am done with patzer AI...

I have used patzer engine too. According to me (just my personal opinion) something is wrong with patzer code. Specially in evaluating section. I didn’t see this problem with your other engines. SFpoly,Raid,Hazard,vanilla (half by you) all of them are very good engines and play and evaluate very well. (Without exaggerating). But about patzer I think the code should be reconsidered completely. It’s just my personal opinion and nothing more.
Best regards and thanks for your constructive activities

Homayoun wrote: Sat Nov 18, 2023 6:51 pm

OrgZ wrote: Sat Nov 18, 2023 4:28 pm

Homayoun wrote: Thu Nov 16, 2023 1:31 pm I thought this engine uses two books by default now. I made a mistake. Anyway, Raid plays very well.
Greetings

Nah, u just load them bin books as Blood.bin and sport.bin been away for a bit Mr H. working on updates, it's a process, trying to make patzer AI less positional and more fun, hv archived that but it alters his strength in long TC.....so back to the drawing board. As for Raid his ready, but will release after am done with patzer AI...

I have used patzer engine too. According to me (just my personal opinion) something is wrong with patzer code. Specially in evaluating section. I didn’t see this problem with your other engines. SFpoly,Raid,Hazard,vanilla (half by you) all of them are very good engines and play and evaluate very well. (Without exaggerating). But about patzer I think the code should be reconsidered completely. It’s just my personal opinion and nothing more.
Best regards and thanks for your constructive activities

Not at all, feel free mr Homayoun.
Am open to suggestions, though am nt sure which version of patzer you are referring to. But what i can say is that the strongest version was 2560i. The high evaluations that it gives, pushes it to fight for a win. But then after a couple of turns of not finding any improvement it gives up n decides to draw, though the position according to other engines might be +/-0.40 (at that tym patzer AI might be +/-0.70) so in longer tym control patzer is capable of proving his evaluation true and wins. But if you like bullet, he won't be able to do it. Its too short and patzer is time engine (needs time to mk things work)

That effect of evals is what AbbyStein uses. Thats y u see her good performance, the major differences are that Abby is more precise and is more selective. I have many positions in my test suite where AbbyStein tops all my engines in finding victories and draws when sides are switched.second is Raid then follows my other side engines. But the current one is different its just giving me problems in blitz, its bullet seems ok but am getting close, though the strong he gets tye more positional he becomes.....

Which time control did you notice the eval effect.....if it was long tym control then you should hv been able to see what he does with that effect.

Though i changed it, trying to mk him more savage (fun)

When we talk about engine’s tournaments or engine vs engine matches it is a different story from when we want an engine for personal analyses and practice. I completely know that patzer has had successes in different tournaments as I study all test sections in this forum and others. But I think if you want patzer be a reliable engine for personal use ,analysing and practice too, you would better works on its playing style and evaluation more. And I think it is related to the source code of engine. I think it’s source code it’s not as mature as your other works. Needs some modification. Unfortunately I am not a chess programmer. Everything I say is just my observations due to my games played with your engine. I don’t know anything about source code and chess programming. May be it’s better to consult with an experienced programmer and engine maker about patzer code and reconsider about it. Although you have shown us ,you are a talented person and all of us enjoy from your fantastic works in chess community.
Best regards

Just a friendly note here, if your derivative (in this case Patzer AI) uses experience file, I highly recommend you to revert whatever patch you implemented that makes it "overestimate" positions (compared to other engines). It is not benefitial neither for the experience file nor for the engine.

And I know this because I had a similar problem with my private derivative so I reverted the changes almost instantly and had a better performance.

Regards.

Homayoun wrote: Mon Nov 13, 2023 8:59 pm

STUDENT_MASTER wrote: Mon Nov 13, 2023 8:41 pm https://pixeldrain.com/u/tHinzwzo

Great work, many thanks and best regards.

+1

Thanks for the Suggestions guys


@ Sedat Canbaz

STUDENT_MASTER wrote: Mon Nov 13, 2023 8:41 pm https://pixeldrain.com/u/tHinzwzo

This version has a bug, its running on an incomplete code, its diffuclt to spot, bcz when it plays its strong but sometimes it messes up evaluations .i discovered it when i was updating it. i'll release the fixed version as v3.1 still testing it.

For those curious about Raid v2.99 2TC
Its derived from sf 031123

viewtopic.php?f=4&t=3778&start=10

Raid v3.1 Blood Sport
https://pixeldrain.com/u/qgPRqGRu

OrgZ wrote: Thu Nov 23, 2023 4:03 am Raid v3.1 Blood Sport
https://pixeldrain.com/u/qgPRqGRu

Many thanks