r_lee

18 ms·

391.
by r_lee 8mo ago
lmao please tell more
392.
by r_lee 8mo ago
I didn't mean to set you off like that? I was just making a counterpoint, because it's easy to just say "have no fear" but I think there is a real place for at least some fear, otherwise why would that feeling exist? if
393.
by r_lee 8mo ago
they're TPUs, same thing as GPUs but specifically for tensor ops.
394.
by r_lee 8mo ago
I would say, just post these kind of rants on a substance and X or something, don't LLM format it, just kind of lay it out and fix whatever typos and let loose it'd be interesting to just hear some thoughts and opinions from someo
395.
by r_lee 8mo ago
So just have no fear? Then when should we have fear? Fear is a natural response, one that in this case is perfectly appropriate imo. I don't fear for my life, but I'm not optimistic at the moment
396.
by r_lee 8mo ago
I fear that there will be a point where it won't be a choice, although a part of me wonders why there is such a hurry to get rid of devs in the first place, as these tech companies have insane margins But I also like to work the way yo
397.
by r_lee 8mo ago
"unexpectedly"
398.
by r_lee 8mo ago
For me Go is like the 80% language. I like TypeScript as well, but Go is just such a reliable workhorse I'd say? it's not "sexy" but it's just satisfying how it's just these simple building blocks that you can
399.
by r_lee 8mo ago
WASM isn't a language you'd want to program with. you can't verify outputs nor is there any proper training data aside from examples and such
400.
by r_lee 8mo ago
it's just someone who thinks degoogling himself has somehow elevated his entire existence to the sky I'm personally not so attached to this idea of Google being evil so I don't really get this at all
401.
by r_lee 8mo ago
I mean, the transformer is basically like a big query engine and the model is the dataset + some logic or whatever it's kind of like that by definition, with the whole Attention stuff etc.
402.
by r_lee 8mo ago
they run on a blend of everything depending on the region from my experience. Hetzner is one of the providers they use
403.
by r_lee 8mo ago
I think its just by nature very engaging, as dudes will go look at other posts and comment (at least the older ones) about their looks etc...
404.
by r_lee 8mo ago
yeah, but dealing with DDoS is easier in terms of DMCA unlike with CDNs because it's you hosting it, not the service provider (this is how Cloudflare avoids DMCA when you cache with them iirc) so if you can just find a good dedicated s
405.
by r_lee 8mo ago
yeah, but I mean more like the old setups where you'd just load a model on a 4090 or something, even with MoE it's a lot more complex and takes more VRAM, right? like it just seems not justifiable for most hobbyists but maybe I&#x
406.
by r_lee 8mo ago
I think that's because there's less local AI usage now since there's all kinds of image models by the big labs, so there's really no rush of people self hosting stable diffusion etc anymore the space moved from Consumer
407.
by r_lee 8mo ago
it's a super cheap "CDN" that runs on Hetzner and random hosts or their colo, it's not as proper as the other ones. for anything DMCA heavy maybe just buying dedicated servers or something instead could work?
408.
by r_lee 8mo ago
like another guy said, it's rhe marketing thing. the open claw thing became a hit sensation in the news etc. and now OpenAI can claim it as theirs. they have infinite money anyways, so might as well buy stuff like that I guess.
409.
by r_lee 8mo ago
I'm also seeing a lot of new rambling in Sonnet 4.6 when compared to 4.5, more markdown slop and pointing out details and things in the context which isn't too useful etc... which then causes increased token usage because you need
410.
by r_lee 8mo ago
it is some form of deterrence, but it's not security you can rely on
411.
by r_lee 8mo ago
I think the sweet spot is to make the initial stuff yourself and then extend or modify somewhat with LLMs it acts as a guide for the LLM too, so it doesn't have to just come up with everything on its own in terms of style or design cho
412.
by r_lee 9mo ago
and if you do a hard refresh on the webapp, it literally takes like 10 seconds for the homepage to load
413.
by r_lee 9mo ago
because it's not easy to identify exactly when to r/w memory accordingly, especially when you'd need to have an LLM decide when and if to do that and to scale it in a way where you don't need a whole custom model loaded
414.
by r_lee 9mo ago
RIP
415.
by r_lee 9mo ago
I would doubt that because books are probably weighed as higher quality and more trustworthy than random Reddit posts Especially if it's unsupervised training
416.
by r_lee 9mo ago
was this actually generated with Claude/GPT? I didn't really notice it at first but on a second read it's full of this crap?
417.
by r_lee 9mo ago
how does that help pay the rent? or how would that get you past resume screening?
418.
by r_lee 9mo ago
but are employers going to be fine with that?
419.
by r_lee 9mo ago
problem today is that there is no "sink" for money to go to when it flows upwards. we have resorted to raising interest rates to curb inflation, but that doesn't fix the problem, it just gives them an alternative income sourc
420.
by r_lee 9mo ago
what makes you think that's actually possible? maybe if you really had the connections and sales experience etc... but also, if that were possible, then why wouldn't prices go down? why would the value of such labor stay so high i

More ›