TalkChess.com

I've just released a new version of my engine.

These are the changes:

Speed optimizations
Tuning
Bug fixing
Don't try null move at low depths
Improved history
Prune moves with bad history
Default DrawScore set to 0
Improved time management
Improved imbalances evaluation

it's 30 elo stronger than 2.0 at short time control. It's the last version with an handcrafted evaluation.
https://sites.google.com/site/pedonechess

Hi Fabio,

shortly ...
BIG THANKS!

Great engine, good king safty, nice games ... also a humen-like-style (I like so much from Wasp).

So, all what you do is right in development of Pedone.
Frank like Pedone a lot ... very important work you do here.

Best
Frank

PS: I hope you will offer Pedone _in additional_ in the future as classic engine in newer versions.
I don't like it if all used NNUE. The personal work by yourself is much more interesting (engine have an own face) as to make the same the others do. Of course NNUE is interesting (no question ... in better english ... there's no doubt about it) but I hope that engine programmers, like you with so many talent, will offer both versions. Please thinking about it, to give you own works in tunings of Pedone (and the works you do here must be definitive good) in the basket isn't ... in my humble opinion ... the right way. Hope I can say that to you, if not ... I like computer chess only and I am harmless. But I will not lose comming soon Pedone versions as classic engine!

Often I wrote Pepito but I means Pedone ...
Pedone and Pepito is a problem for my brain I think.

It's really important that the TOP chess programmers give the _right_ signal / message ...

NNUE is interesting = yes!
BUT to hold the own work and using NNUE max. in additional is maybe more interesting!!

Fabio, if you and different other _programmer VIPs_ goes here the right way (in my opinion) it will be a better message to the group of the other chess programmers. I am quite sure that more and more persons lost interest on programs, not to find on place 1-5 in rating systems, if NNUE used only. For the most person, I have personal contacts, is the work by the programmers in self-work tunings much more interesting.

The own face, the completly own style ...
Should be never give-up for boring auto-tunings!

Best
Frank

Thanks, Fabio.

Neural nets give a better evaluation than an handcrafted one. I'm going to try to implement my own neural network with my own training method so the evaluations will be different from other implementations. Improve the evaluation is very difficult and if I can replace it with a better one I will be happy although training a neural net is a boring part of the development of a chess engine.

Thanks Fabio for this new release !

thank u for adding android version everytime

Is it true that Pedone is now using NNUE?

Fabio Gobbato wrote: ↑Thu Aug 20, 2020 8:34 pm I've just released a new version of my engine.

Hi Fabio. Is it true you have trained your own NNUE network to go along with SF's / CFish's optimized implementation? Congrats if so.

Yes Pedone is now using NNUE. I have implemented my own network with included castle rights and with independent weights for the both sides.
For the training I have used 14 depth search score of Pedone with traditional evaluation and I have used a simple SGD to train the net.
I'm not an expert on NN but I have found on the net a lot of material that have helped the development.
I also have a discussion with Marco Belli and we have helped each other in implementing NNUE.