Are neural nets (the weights file) copyrightable?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Ckappe
Posts: 81
Joined: Sun Feb 14, 2021 11:50 am
Full name: Rütger Andersen

Re: Are neural nets (the weights file) copyrightable?

Post by Ckappe »

dkappe wrote: Mon Feb 22, 2021 6:44 pm So since a0lite won’t even work without a net and is under the MIT license, that same network would be under the MIT license or whatever onerous license I put on it? Also, how do you come to a conclusion that all networks that could possibly be used with Minic would have to be under the GPL? Reading the FAQ, the getting started doc and the GPLv3, I don’t see anything that supports your theory.
By logic, the network (if seen as code) could apply any licensing if it is to be used separately and independently from other code. if it has to be used as part of a controlling application it will by the same rationale have to be under a license compatible with the license of the program that calls this "subroutine", otherwise, the licenses would be incompatible.
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Are neural nets (the weights file) copyrightable?

Post by dkappe »

Ckappe wrote: Mon Feb 22, 2021 7:06 pm
dkappe wrote: Mon Feb 22, 2021 6:44 pm So since a0lite won’t even work without a net and is under the MIT license, that same network would be under the MIT license or whatever onerous license I put on it? Also, how do you come to a conclusion that all networks that could possibly be used with Minic would have to be under the GPL? Reading the FAQ, the getting started doc and the GPLv3, I don’t see anything that supports your theory.
By logic, the network (if seen as code) could apply any licensing if it is to be used separately and independently from other code. if it has to be used as part of a controlling application it will by the same rationale have to be under a license compatible with the license of the program that calls this "subroutine", otherwise, the licenses would be incompatible.
See the GPL FAQ. Since the neural net weights cannot be executed directly on a computer but have to be interpreted by another program, it’s not covered by the interpreter’s license (have we been here before, again and again?). I won’t quote the GPL FAQ again. It’s getting tiresome.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Ckappe
Posts: 81
Joined: Sun Feb 14, 2021 11:50 am
Full name: Rütger Andersen

Re: Are neural nets (the weights file) copyrightable?

Post by Ckappe »

hgm wrote: Mon Feb 22, 2021 6:45 pm
Michel wrote: Mon Feb 22, 2021 5:42 pmPeople claim that the because the net is not directly executed by the CPU but instead is interpreted by SF it is exempt from the GPL. I do not buy this. I consider the method by which code is translated into actions irrelevant.
By that reasoning GPL'ed interpreters (or 'just-in-time compilers) could not be used to run non-GPL programs of their target language. We know that is not the case.
I don't see equivalence here.. A platform, runtime, or OS is not the same as this case which is more of an application program using a sub-routine that is supplied as a network.

The chess engine is an application not a platform / OS.
Michel
Posts: 2272
Joined: Mon Sep 29, 2008 1:50 am

Re: Are neural nets (the weights file) copyrightable?

Post by Michel »

hgm wrote: Mon Feb 22, 2021 6:45 pm
Michel wrote: Mon Feb 22, 2021 5:42 pmPeople claim that the because the net is not directly executed by the CPU but instead is interpreted by SF it is exempt from the GPL. I do not buy this. I consider the method by which code is translated into actions irrelevant.
By that reasoning GPL'ed interpreters (or 'just-in-time compilers) could not be used to run non-GPL programs of their target language. We know that is not the case.
The distinction I make is that in SF's case the interpreter and the interpreted program are not independent ("they are linked"). SF brings search and the interpreted program brings the evaluation.

For me the fact that a part of a larger program is interpreted versus it being directly executed on the CPU is irrelevant.

Anyway such things are not decided on Talkchess.
Ideas=science. Simplification=engineering.
Without ideas there is nothing to simplify.
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Are neural nets (the weights file) copyrightable?

Post by dkappe »

Michel wrote: Mon Feb 22, 2021 7:22 pm Anyway such things are not decided on Talkchess.
Finally we can agree! :D
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Are neural nets (the weights file) copyrightable?

Post by gonzochess75 »

Ckappe wrote: Mon Feb 22, 2021 7:21 pm
hgm wrote: Mon Feb 22, 2021 6:45 pm
Michel wrote: Mon Feb 22, 2021 5:42 pmPeople claim that the because the net is not directly executed by the CPU but instead is interpreted by SF it is exempt from the GPL. I do not buy this. I consider the method by which code is translated into actions irrelevant.
By that reasoning GPL'ed interpreters (or 'just-in-time compilers) could not be used to run non-GPL programs of their target language. We know that is not the case.
I don't see equivalence here.. A platform, runtime, or OS is not the same as this case which is more of an application program using a sub-routine that is supplied as a network.

The chess engine is an application not a platform / OS.
Hi Ckappe, this thread is not going to go anywhere arguing with people who've devoted themselves to tying themselves in knots trying to find a way to violate others copyright. I've argued with them enough to try and spell away their silly non-sequiturs and so on probably for naught. Anyway, - you probably already know this - but the 'OS' designation wouldn't automatically absolve anyone of linking a proprietary module to GPL'd code for instance.

You can see from none other than Linus Torvalds here: https://ipfs.io/ipfs/QmdA5WkDNALetBn4iF ... dules.html
Does that mean that any kernel module is automatically not a derived work?
HELL NO! It has nothing to do with modules per se, except that non-modules
clearly are derived works (if they are so central to the kernel that you
can't load them as a module, they are clearly derived works just by virtue
of being very intimate - and because the GPL expressly mentions linking).
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Are neural nets (the weights file) copyrightable?

Post by dkappe »

gonzochess75 wrote: Mon Feb 22, 2021 7:35 pm Hi Ckappe, this thread is not going to go anywhere arguing with people who've devoted themselves to tying themselves in knots trying to find a way to violate others copyright. I've argued with them enough to try and spell away their silly non-sequiturs and so on probably for naught. Anyway, - you probably already know this - but the 'OS' designation wouldn't automatically absolve anyone of linking a proprietary module to GPL'd code for instance.
Whose copyright have I violated?
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Ckappe
Posts: 81
Joined: Sun Feb 14, 2021 11:50 am
Full name: Rütger Andersen

Re: Are neural nets (the weights file) copyrightable?

Post by Ckappe »

dkappe wrote: Mon Feb 22, 2021 7:14 pm See the GPL FAQ. Since the neural net weights cannot be executed directly on a computer but have to be interpreted by another program, it’s not covered by the interpreter’s license (have we been here before, again and again?). I won’t quote the GPL FAQ again. It’s getting tiresome.
I don't think the GPL FAQ supports your conclusions. I think it is a false assumption to categorize applications using an NN as "interpreters" simply because some parts of the code "interpret"s the NNs. it's a much tighter coupling between an NN and the main program than a generic runtime-environment, which is referred to with interpreters in GPL.

If the interpreter ONLY interprets the NN then I would agree with your reasoning. But in this case it does so much more than just "interpreting" an NN so you cant really classify SF etc. as an "interpreter"..

From the GPL faq:

"If a programming language interpreter is released under the GPL, does that mean programs written to be interpreted by it must be under GPL-compatible licenses? (#IfInterpreterIsGPL)

When the interpreter just interprets a language, the answer is no. The interpreted program, to the interpreter, is just data; a free software license like the GPL, based on copyright law, cannot limit what data you use the interpreter on. You can run it on any data (interpreted program), any way you like, and there are no requirements about licensing that data to anyone."
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Are neural nets (the weights file) copyrightable?

Post by gonzochess75 »

dkappe wrote: Mon Feb 22, 2021 7:42 pm
gonzochess75 wrote: Mon Feb 22, 2021 7:35 pm Hi Ckappe, this thread is not going to go anywhere arguing with people who've devoted themselves to tying themselves in knots trying to find a way to violate others copyright. I've argued with them enough to try and spell away their silly non-sequiturs and so on probably for naught. Anyway, - you probably already know this - but the 'OS' designation wouldn't automatically absolve anyone of linking a proprietary module to GPL'd code for instance.
Whose copyright have I violated?
None that I am aware of. Do you have something you want to get off your chest?

What I am aware of is you trying desperately - in this very thread - to find a way with tortured readings, illogic, and non-sequiturs to justify and apologize the ability of others to do so by finding non-existent loopholes.
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Are neural nets (the weights file) copyrightable?

Post by gonzochess75 »

gonzochess75 wrote: Mon Feb 22, 2021 7:53 pm
dkappe wrote: Mon Feb 22, 2021 7:42 pm
gonzochess75 wrote: Mon Feb 22, 2021 7:35 pm Hi Ckappe, this thread is not going to go anywhere arguing with people who've devoted themselves to tying themselves in knots trying to find a way to violate others copyright. I've argued with them enough to try and spell away their silly non-sequiturs and so on probably for naught. Anyway, - you probably already know this - but the 'OS' designation wouldn't automatically absolve anyone of linking a proprietary module to GPL'd code for instance.
Whose copyright have I violated?
None that I am aware of. Do you have something you want to get off your chest?

What I am aware of is you trying desperately - in this very thread - to find a way with tortured readings, illogic, and non-sequiturs to justify and apologize the ability of others to do so by finding non-existent loopholes.
And to be very very clear... as I said from the get go I don’t think any copyright has been violated because these net files are not copyrightable. Lucky ducks!