Simon Richter <[email protected]> wrote on 25/08/2026 at 05:27:32+0200:
> Hi Wookey, > > On 8/25/26 8:49 AM, Wookey wrote: > >> Having read the various options and some of the discussion on -vote I >> am surprised that none of the options we have to vote on seem to >> distinguish between LLMs where we have all the parts and can run them >> ourselves, and proprietary LLMs run remotely over which we have no >> control, reproducibility options, security/privacy insight, etc. > > We have the necessary tools in Debian, but ingesting the entirety of > the Debian source code and mailing list archive, which is all that we > reasonably have access to in a reproducible way, would not be > sufficient for anything but a fuzzy code search. > > The training set would be too small to argue that copyrights are > sufficiently diluted that we can ignore licenses (which is a dangerous > position that we should not adopt), so code generation is out. Code generation is not out, there is nothing very specific to Debian, and you can take a strong model and add a RAG layer for Debian specific data, documentation etc. I think the reason there is no dedicated ballot options is that the GR has been open as "Ban LLM contributions", which set the goalpost to LLMs as a whole. Also, having the compute resources to run propers models for debian's purposes is not trivial. But maybe we coul find support with a hoster/company eager to lend us some machines with GPUs and then it becomes openrouter+some infra bits. -- PEB
signature.asc
Description: PGP signature
