*First release* Giraffe, a new engine based on deep learning

[fern]

I used Shredder settled to use Winboard.
I have no idea how the program can use that .net thing you mention. I suppose Sh simply goes and take the .exe.
As you see, I am not an expert, just a writer, an user and a mediocre chess player.
Do not bother with me, there are lot of people here that DO know how to test your program, as Frank no doubt is

Good luck and good night as someone said

Fern

[matthewlai]

I used Shredder settled to use Winboard.
I have no idea how the program can use that .net thing you mention. I suppose Sh simply goes and take the .exe.
As you see, I am not an expert, just a writer, an user and a mediocre chess player.
Do not bother with me, there are lot of people here that DO know how to test your program, as Frank no doubt is

Good luck and good night as someone said

Fern

Thanks. Yeah I have never used Shredder so I have no idea. Usually the setting is called "working directory". It needs to be set to the directory the .net file is in.

Good night to you as well.

[fern]

I detected the problem: your progran run at 64 bit, my comp at 32......

Well....

[matthewlai]

I detected the problem: your progran run at 64 bit, my comp at 32......

Well....

Ah! Yes, that would be a problem. It's 64-bit only unfortunately.

[supersharp77]

Nice engine but a windows 32 edition would be greatly appreciated for extensive testing...Brainless is a fascinating engine w32 (once you work thru the flaws/kinks)......Thx AR

[matthewlai]

Nice engine but a windows 32 edition would be greatly appreciated for extensive testing...Brainless is a fascinating engine w32 (once you work thru the flaws/kinks)......Thx AR

Thanks!

Aaah Brainless. It's been so long I almost forgot about it .

I'll try to make a 32-bit version for the next update. It will probably require some tweaking to work in 32-bit.

All of my machines (and all the machines I have access to) have been 64-bit for so many years that I always forget some people are still on 32-bit.

[gerold]

Works good in Arena. King safety will have to be worked on. Leaves the King open to attack and King will also go out in the open and attack.
Not to hard to beat the engine at this time. Fun trying it out.

Best,
Gerold.

[matthewlai]

Works good in Arena. King safety will have to be worked on. Leaves the King open to attack and King will also go out in the open and attack.
Not to hard to beat the engine at this time. Fun trying it out.

Best,
Gerold.

Glad you had fun . Hopefully next update will be more challenging!

The cool thing about this engine is that it actually figured almost all of the evaluation stuff out by itself. I didn't give it any chess knowledge at all, besides basic values of pieces.

King safety is bad probably because the evaluation function is the same throughout the game, so it doesn't know kings aren't supposed to go out until end game. That should be improved in the next update.

[kinderchocolate]

Good work. Do you think you can open your thesis because I'm very interested about it? I've tried neutral network myself, but I'm picking up deep learning recently.

[matthewlai]

Good work. Do you think you can open your thesis because I'm very interested about it? I've tried neutral network myself, but I'm picking up deep learning recently.

That's the plan. It hasn't been written yet .

Deep learning is a lot of fun! It's also a very exciting field academically, with major discoveries still being made every few months. There aren't many other fields moving that fast.