I guess I can delete my 1.6TB stash of models now.
/s
I guess I can delete my 1.6TB stash of models now.
/s
Don’t, given the unpredictable track record of the current US admin on subjects taken a stance on ![]()
As strong anti-AI proponent: please do! ![]()
I forgot the /s ![]()
Software engineers use local LLMs very differently then what the vast public use them for.
I have a measurable speedup of 4x more productivity using my local setup over 9 months now compared to the year before.
…While still reviewing, understanding, and owning every line of code I commit. YOLO has no place in software engineering.
I hate that my public code was used to train this crap. But it’s needed to stay up to date and software exploit preventions.
For example, previously Red Teams were all serial. One exploit attempt at a time. Now, you can have 20 agents attempting parallel review and escapes at the same time.
Knowing how the orange administration operates, expect this to change randomly every 2-4 weeks.
As a strong anti-self-defeating choice proponent: please dont.
The truth is that no matter what this admin says, it would be extremely hard to regulate any open source, open weight models, because you would need to censor the entire internet, and historically, that does not work. People would just share the files through file sharing networks, and the whole thing would ultimately just lead to more bypasses being built for everyone to continue business as usual.
We have seen this happen in other industries and we’ve also seen the censorship attempted in other countries. Mido scenario has ever worked. So whether or not you want to delete those models depends entirely on how useful those models are to you today and how useful they will be to you in the future.