Tim

I just realised this is the much more useful link: https://github.com/p-e-w/heretic?tab=readme-ov-file

I can see at least one -heretic version of a Qwen3.5 model on Huggingface already; can’t vouch for quality though.

Keep an eye on this: https://huggingface.co/heretic-org

I used to use a -heretic abliterated version of gpt-oss-120b, not for any creative reasons but just to reduce the amount of wasted tokens in its thinking, with good results.

(You can turn off thinking mode with the new Qwen models btw - how you do it will depend on how you’re hosting it, but basically it’s a flag to the chat template. It won’t remove the safety guidelines, but it will stop it telling you all about its internal monologue ;).)

Advanced Vector Extensions instruction set; introduced with Sandy Bridge in 2011, but not included in Pentium/Celeron branded processors even after then for reasons best known only to Intel.

Mongo is the application that has most irritated me by requiring it, but I doubt it’s the only one.

Just throwing this in here as another thing to consider - instruction set. From a quick check (so I’m happy to be told I’m wrong) the Celeron & Pentium options don’t support AVX. That means some stuff - and I’m giving a hard stare at MongoDB here, but there will be others - is not going to run, or at best you’re going to be either stuck with old versions or recompiling yourself from source.

(I don’t know if any of your apps require Mongo or AVX, but I was bitten by this in the past and it was one of the main reasons I eventually upgraded one of my small clusters.)