@HSVSphere Honestly not having flexibility is very annoying but setting up my flake so I can set up both of my systems seperately from one file that I control via git is unbeatable
@jsuarez Dont tell anyone but I'm building a pure RL nlp agent. Supervised learning is the inferior mode for learning anything. With RL you can change the context form MLM to autoregressive easily as long as you use a good optimizer.
One reason I've been hesitant to work on alectors (or telestes more specifically, since alectors are built on top), is that I know in my heart that I want to do it in Nim rather than Python.
The speed of compiled binaries and the flexibility of macros make it so appealing.
@tauraamuix@BrodieOnLinux He doesnt need to "instruct" anyone. The mere fact that he maintains that this is a moral "us vs them" issue means that someone will consider that as the moral side they need to do something about it.
If you yell "ALL WITCHES ARE EVIL" and claim someone is a witch
@dhh@dhh I dont know if you're further interested in my viewpoint on the subject, but in any case I wrote a post about it (had been meaning to for a lon time)
https://t.co/CffQbxRLkM
@dhh hell yeah, different opinion, comparable mode of thinking. My approach is that I don't want to enable a Minix situation.
"[the Law] is the collective organization of the individual right to lawful defense." per Bastiat. I see the GPL as that essentially.
https://t.co/slajDEWZmR