MIT is more liberal, so it can be included in a GPL'd project. Someone receiving the GPL'd project's source code may then rip out the part of the code that is under the MIT license and release that without being restricted by the GPLv3. That's fine, because the copyright holder permitted it by releasing that part of the code under the MIT license.
So there is no real problem that I can see, only a potential problem in case modifications are added to it released only under the GPLv3.
It is fine(as a copyright owner) to release anything under more "free"(compatible) license which may include parts of GPL code. Just the whole thing then becomes GPL to everyone else unless they remove the parts covered by GPL.
I have made a PR for the nodchip repo to change the license text. I have also filed an issue in the pytorch trainer that the code that is taken from the nodchip trainer needs an independent implementation. Thank you for your contribution.
dangi12012 wrote:No one wants to touch anything you have posted. That proves you now have negative reputations since everyone knows already you are a forum troll.
Maybe you copied your stockfish commits from someone else too?
I will look into that.
Sopel wrote: ↑Sun Feb 28, 2021 1:09 pm
I have made a PR for the nodchip repo to change the license text. I have also filed an issue in the pytorch trainer that the code that is taken from the nodchip trainer needs an independent implementation. Thank you for your contribution.
I may be confused about the details now, but if the pytorch trainer is GPLv3 and the current implementation includes code licensed under the MIT license, then everything seems fine.
I've accidentally discovered that, 9 hours ago, Martin made 11 commits to the nnue-trainer repo and, in particular, replaced the dataloader with a new one, so the repo seems free from GPLed code and thus legitimately under MIT now. Thanks a lot to Martin!