For open-weights models, censorship removal is now a "solved" problem. If you wait a few days after a new model release, someone will have made a heretic ( https://github.com/p-e-w/heretic ) version with the censorship removed, so in a way the only use for censorship now is to avoid lawsuits, not reduce improper usage.
Any time I've tried an "abliterated" model, heretic or other, it has always damaged the capabilities of the original model and will still often refuse or produce garbage at a lot of "unsafe" requests.
Abliteration can't teach the model something that wasn't in pre-training, it's just fixing refusals from post-training. I don't find the delta to be that big in practice and it really depends on what you're doing with the models anyway. If your primary usecase is sexy roleplay I think the loss of absolute capability is probably worth the abliteration, for malware research it's probably better to just jailbreak.
I've mostly found that finetunes and abliterations are of limited use but that's recently changed for me. My default model for the past week or so has been a Qwen 3.6 tuned on Opus 4.7, it's definitely a bit worse than the base Qwen in terms of precision and "intelligence", but it MORE than makes up for it in response style. Way easier to get it to write things that I want to read, it's way more terse, way fewer emoji. Best local rubber duck by far.
There are many abliterations which work quite well. Older techniques do suffer from quality issues, but more recent ones do a much better job. In particular, the older approaches did poorly on MoE models.
Another likely problem you're running into: the problems with older techniques compound with quantization. Anything less than 5-bit quant is going to give you some pretty sketchy outputs, in my experience.
The problem is the heretic and abliteration versions are dog shit quality compared to the non-edited versions and much more likely to hallucinate.
AFAIK abliteration without quality reduction isn’t even possible without some quality reduction, even if it’s marginal. All the benchmarks reflect this.
Why not try to find a successor instead of archiving the repo and forbidding the use of the name? I'm sure with a 3.8k stars repo you'll find competent people willing to continue the work.
Sometimes you want to hang things to your wall, and be done with it.
I'd personally do the same. I wouldn't want to be bothered by the future maintainers' choices and get feedback/flak for it. It's a well-known and well-respected way to cycle the name with a "-ng" or "-nx" prefix to signal that this is the newer project with a different set of maintainers.
Being MIT, while is not my favorite license, doesn't give free license to grab and run with things.
Honestly, in my eyes, 3.8K or 38K stars mean nothing, because Open Source is not about you [0], to begin with.
It is reasonable to ask for a follow-up project/fork to take a different name. Naming your project, e. G., pgbackrest-ng, does not sound too onerous of a requirement and clearly communicates to users that maintainers have changed (see also paperless ng/ngx as good examples of such a change).
Finding a successor is also not easy nor cheap (in regards to time).
You'll also find plenty of potential malware injectors too, and who would want the responsibility of trying to vet a successor and have to work out the difference?
Because you will attract people who will want to take advantage of the trust these 3.8k stars signal to some people, for example, by means of supply chain attacks.
The Apache Foundation used to help with this sort of governance problem didn't it? Thugh maybe pgbackrest isn't quite big and official enough to be the kind of software which Apache takes on, and one certainly hears (increasing?) grumbles about Apache's stewardship.
There's no way to know if a new maintainer will live up to whatever standards they've kept to date. Archiving should be the default decision, unless there's formal and elaborate handover.
A maintainer that is mainly motivated by the 3.8k stars aspect is probably not the person you want. Working on critical OSS software is fun until it's not, especially when you are not paid for that work.
Those people can just as easily fork it and make a new name then. Otherwise you end up with situations where it's actually an entirely new thing under new developers under the same name. Even riskier in the age of the "AI clean rewrite"
The latest Qwen3.6 model is very impressive for its size. Get an RTX 3090 and go to https://www.reddit.com/r/LocalLLaMA/ to see the latest news on how to run models locally. Totally fine for coding.
I completely understand why measuring the length of coastlines is not possible but surely measuring a trail should be doable quite easily, you could simply use a gps tracker and it would be precise enough.
I think the idea is to show we don't know the exact lengths of any paths that aren't constructed from our handful of mathematically known curves and must approximate using them instead.
If you measure with GPS coordinates, you still run into the same problem. The number of points plotted onto a curve affects the result, and then you are possibly also adding more error than you'd have compared to tracing from aerial photos.
