1,682 karma · joined March 21, 2016
Great men I admire: - Jimi Hendrix - Claude Shannon - Carl Sagan
Might be. Or might not be. Who knows? Let's leave it up to random chance and see what happens! No point in trying to understand this phenomenon right?
What a great victory!
Aka, politics as usual.
3. Train for only 1 epoch - interesting, any known rationale here?
5. I just read somewhere else that someone got good results from mixing their custom model with the original (60/40 in their case) - good to hear some more anecdotes that this is pretty effective. Especially the further training after merging, sounds promising!
I've also been using kohya_ss for training LoRAs so great to hear it works for you for models as well. On your point about the inference tricks, definitely noted but I did notice that you can feed some params (# of samples, negative embeddings, etc) to the sample images generated during training (check the textarea placeholder text). Still not going to have all usual the tricks but it'll get you a little closer.
I've never even thought about the social aspect of being a react dev - is there one? Frontend sure, but react specifically? It's just a tool, nothing to get ideological or bent out of shape over.
Couldn't give a shit what Abramov says beyond what is useful in my day-to-day. There are probably a lot of junior devs who are starstruck but I've never seen any hint of ostracization (lol). There are tons of junior web devs (and more everyday) though so I can understand how one could form that perception.
Yes to less celebrities and more problem solving.
Hooks are fantastic and so much more composable than the class methods. I use 100% hooks and functional components in my code nowadays and it has never been easier to build and understand my UIs - especially those with non-trivial logic and state. Never had a problem with them being "simple on the surface" but ymmv I guess.
Best just to dive in if you're interested IMO. Otherwise you'll get lost in all the new jargon and ideas. Great place to start is the A1111 repo, lot of community resources available and batteries included.
And vice versa, which is the exciting part to me - only a matter of time!
This is basically my dream for local AI... locals models trained on my own data/code/styles. Even if they're slow, as long as they work (V/RAM) and are of high enough quality then I'm happy to wait!
We've been hearing these same arguments since the beginning... If nothing else Bitcoin has shown impressive sticking power, given all of its stumbles and uphill battles.
What would a modern day Nuremberg trial look like?