make3

9 ms·

241.
by make3 9mo ago
People on HN are not regular users in any way, shape or form. It's just the "cognitive load" UX idea, with extremely non-technical people having extremely low limits before they decide to never try again, or just feel intimid
242.
by make3 10mo ago
The advantages of having a single big memory per gpu are not as big in a data center where you can just shard things between machines and use the very fast interconnect, saturating the much faster compute cores of a non Apple GPU from Nvidi
243.
by make3 10mo ago
AI is threatening to remove humans from the whole equation except the very top. AI is an existential threat (not in the Terminator sense). Especially for art, I'm an AI researcher myself (in bio for health), but I think that ppl are co
244.
by make3 10mo ago
I think it's not uncommon to ask an LLM for the bibtex for a paper you know about, & it might mess it up, but that doesn't feel like a fireable offense
245.
by make3 10mo ago
This interpretation would have been ok for old generation models without search tools enabled and without reliable tool use and reasoning. Modern LLMs can actually look up the existence of papers with web search, and with reasoning, one can
246.
by make3 10mo ago
Isn't this mostly a set of citation typos? To me this mostly calls for better bibtex checking, writing and checking bibtex is super annoying
247.
by make3 10mo ago
Architecture difference wrt vanilla transformers and between modern transformers are a tiny part of what makes a model nowadays
248.
by make3 10mo ago
I suspect they will keep doing this until they have a substantially better model than the competition. Sharing methods to look good & allow the field to help you keep up with the big guys is easy. I'll be impressed if they keep pub
249.
by make3 10mo ago
yes, they turned off all energy economy measures when benchmarking software activity was detected, which completely broke the point of the benchmarks because your phone is useless if it's very fast but the battery lasts one hour
250.
by make3 10mo ago
I would assume that huge amount is spent in frontier models just making the models nicer to interact with, as it is likely one of the main things that drives user engagement.
251.
by make3 11mo ago
this is such a bad way to evaluate a metric. we want something that can distinguish small percentage changes in affordability for the people who live there
252.
by make3 11mo ago
Sidenote, it's interesting how the term "virtue signaling" is arguably objectively an individualistic right-wing dog whistle these days. I would argue that it is being used all over the media to complain about anyone showing
253.
by make3 11mo ago
That's one of the things that users think they want, but use the product 30x when it's not actually that way, a bit like follow-only mode by default on Twitter etc.
254.
by make3 11mo ago
"with the approval of the intended recipient" makes it like breaking into someone's home with their consent
255.
by make3 11mo ago
I don't know why you assume that people are saying negative things about PyTorch. Most people by very very far really love it.
256.
by make3 11mo ago
In the US, one has to remember that megacorps usually end up winning, as people mostly only care about money and blindly supporting corporations is seen as pro-economy. There are also no limits on political financial contributions, and peop
257.
by make3 1y ago
yes but that could mean a layer per model
258.
by make3 1y ago
these other models are likely much smaller
259.
by make3 1y ago
the models are huge, so not a single (latest gen) one can fit on a single GPU. It's likely that these are small unpopular (non flagship) models, or that they only pack eg one layer of each model.
260.
by make3 1y ago
aren't all of these multimodal LLM approaches, just open vs closed ones
261.
by make3 1y ago
they're hiring people for that right now if you look at their LinkedIn posts
262.
by make3 1y ago
gpt 5 is not a dud though
263.
by make3 1y ago
the "conscious emerges from quantum superposition" idea is controversial because it's thinly veiled mysticism
264.
by make3 1y ago
huggingface scares me in that regards
265.
by make3 1y ago
isn't that just instruction fine tuning and rlhf inducing style & deference? why is that surprising
266.
by make3 1y ago
it's called Roblox and it's bigger than Flash ever was
267.
by make3 1y ago
no one is saying rag is dead, you're never going to put the whole Internet in the context of the model, & the more you put the more expensive it is.
268.
by make3 1y ago
you're the problem
269.
by make3 1y ago
is there a more VC oriented site, this one is software engineering oriented
270.
by make3 1y ago
figuring out how to turn on brain plasticity and motivation networks, and just kill anxiety, would make life quite interesting. it's too bad that improvements in neuroscience are the absolute most insane Pandora's box, with dictat

More ›