bavell

10 ms·

121.
by bavell 1y ago
Sounds like a good way of tackling this - thanks for sharing!
122.
by bavell 1y ago
Iirc abliteration (ablation?) can be done without "training" and is pretty quick. It finds the individual weights related to the concept you want to ablate, and modifies those weights to "deactivate" them. Precision brai
123.
by bavell 1y ago
Most minimum/recommended game specs reference mainstream gaming CPU/GPUs, and most gamers know the strength of their own hardware relative to mainstream components. If you're a very casual/young/inexperienced gamer
124.
by bavell 1y ago
Correlation is not causation.
125.
by bavell 1y ago
I play all my steam games on Linux/proton, zero issues for years now.
126.
by bavell 1y ago
> Current video games are designed around streamlining content. As a player, your job is to extract all content from an area before going to the next. Wrong right from the outset. Some games are designed around content and "extrac
127.
by bavell 1y ago
Has anyone said they actually want this? I wouldn't, especially for traditional folktales.
128.
by bavell 1y ago
You basically replied to GP's eloquent and nuanced post with, "nah bro trust me, purple is actually blue!"
129.
by bavell 1y ago
China has a demographic time-bomb about to go off in the next decade or two. We'll see if they can survive it.
130.
by bavell 1y ago
Nah, if you have an important application this is very low cost for adding tons of insight into how your app is running.
131.
by bavell 1y ago
It's fairly common for artists to release a single or two from an upcoming album. Some artists release more singles than they do albums.
132.
by bavell 1y ago
IDK about "efficiently" but we've been able to run llms locally with AMD for 1.5-2 years now
133.
by bavell 1y ago
Chroot jail?
134.
by bavell 1y ago
> how do you accept/reject and rollback changes? Git
135.
by bavell 1y ago
Great read and insights! I learned a few things and found myself nodding along most of the article.
136.
by bavell 1y ago
> I think OpenAI may have taken the medal for best available open weight model back from the Chinese AI labs. IMO, the "I think..." bit could be ambiguous and read as, "In my opinion, OpenAI may have...". I agree with
137.
by bavell 1y ago
Did you mean to type equality? As in, "everyone on equal footing"? Otherwise, I'm not sure how to parse your statement.
138.
by bavell 1y ago
I mean, there's always torrents
139.
by bavell 1y ago
I'd love to try this on my data set - what approach/tools/models did you use for fine-tuning?
140.
by bavell 1y ago
There are a ton of ways to do this that haven't been tried yet.
141.
by bavell 1y ago
Unfortunately, it was predictable given the other "choices"
142.
by bavell 1y ago
> I've basically wasted the morning on Claude Code when I should've just been doing it all myself. Welcome to the machine https://www.youtube.com/watch?v=tBvAxSx0nAM&t=45s
143.
by bavell 1y ago
I'm on the basic $20/mo sub and only ran into token cap limitations in the first few days of using Claude Code (now 2-3 weeks in) before I started being more aggressive about clearing the context. Long contexts will eat up tokens
144.
by bavell 1y ago
The genius bureaucrat probably did as much research as you did and came to the same conclusion.
145.
by bavell 1y ago
Yes, incredible productivity gains from using a single language in frontend and backend.
146.
by bavell 1y ago
They're referring to the training data being garbage, not the diffusion process.
147.
by bavell 1y ago
I've been wondering this for literally years now...
148.
by bavell 1y ago
Fantastic comparisons! Great to see the limitations of the latest models.
149.
by bavell 1y ago
Great Q&A's, thank you for taking the time to answer! Sounds like a great way to handle the complexities of business reality.
150.
by bavell 1y ago
First time hearing about oRPC, never heard of or used ts-rest and I'm a big fan of tRPC. Is the switch worth the time and energy?

More ›