Indeed, thanks for pointing this out and the links. With the excitement I misread that it was an MR from the fork to the main project. I don’t think I’m able to

Indeed, thanks for pointing this out and the links. With the excitement I misread that it was an MR from the fork to the main project. I don’t think I’m able to fix the title though. I find it quite exciting to read some results in an effort to understand if TurboQuant main ideas can be applied to model weights. There are other similar projects, so we’ll see, but it seems some of this fork results look promising.