gwern

29 ms·

451.
by gwern 2y ago
What people are discovering with the latest models is that often their errors are due to entirely reasonable choices and assumptions... which happen to be wrong in your specific case. They call a library you don't have installed, or so
452.
by gwern 2y ago
I think OP makes a good case that the cause of the fire was probably the shortsighted poisoner, child rapist, human/dog murderer, embezzler, thief, and bomber who apparently had grudges against the captain who mysteriously died shortly
453.
by gwern 2y ago
Their creative-writing dataset seems to have been literally written by ChatGPT: https://www.reddit.com/r/LocalLLaMA/comments/1i1a88y/minimax...
454.
by gwern 2y ago
Duplicate: https://news.ycombinator.com/item?id=42718166
455.
by gwern 2y ago
> I don't think you're going to find an LLM with a large enough context window to have a meaningfully involving story spanning multiple sessions. First, you don't need much of any context window because you can finetune t
456.
by gwern 2y ago
Why is that not just 'transclusion' https://en.wikipedia.org/wiki/Transclusion ? Except a special-case of transclusion: high latency and compile-time only. As opposed to doing it with JS dynamically; if Pinbo
457.
by gwern 2y ago
You are observing "flattened logits" https://arxiv.org/pdf/2303.08774#page=12&org=openai . The entropy of Chat GPT (as well as all other generative models which have been 'tuned' using RLHF, in
458.
by gwern 2y ago
OP is an example of how AI-generated images are usually clutter. Not only do the images not add anything meaningful to the text, and arbitrary parts of the images could be deleted or randomized without affecting the reader's understand
459.
by gwern 2y ago
The bank account anchoring bias may be just the base model predicting as best as it can, as OP described earlier. Imagine a random Internet page where you read "There is V in my X. The Y of Z A, in Bs, is _." What do you expect? V
460.
by gwern 2y ago
> I'll cherry pick just one example from this article. The author in one sentence dismisses the entire field of psychoanalysis. Properly. The issues with psychoanalysis are extensive, well known, and he has discussed them in the pas
461.
by gwern 2y ago
The Waymo criticisms are absurd to the point of dishonesty. He criticizes a Waymo for... not pulling out fast enough around a truck, or for human criminals vandalizing them? Oh no, once some Waymos did a weird thing where they honked for a
462.
by gwern 2y ago
Hm... I was definitely under the impression that it is generating SVGs natively, and that was consistent with its output and its recent upgrades like good text rendering, and I'm fairly sure I've said as much to the CEO and not be
463.
by gwern 2y ago
No, I actually was referring to their native vector AI image generator, not their vector izer - although the vectorizer was better than any other we found, and that's why we were using it to convert the Midjourney PNG dropcaps into SV
464.
by gwern 2y ago
Have you looked at https://www.recraft.ai/ recently? The image quality of their vector outputs seems to have gotten quite good, although you obviously still wouldn't want to try to generate densely textured or photogra
465.
by gwern 2y ago
I am flattered to hear all that. I think Said is a lot closer to the '1px is broken' than me, but also that it's not really that important. How many websites have dropcaps at all? They are fun to have, but very optional. And
466.
by gwern 2y ago
Oh, I assumed the image gen at least was running on the fancy remote GPU server. But it being on-device doesn't prove it's not a DALL-E 3. Just like GPT-3 & GPT-4 & Llamas come in many sizes, some of which are (probably) f
467.
by gwern 2y ago
It's astonishing to see Apple settle for DALL-E 3 (or worse?! these remind me of the Bing samples) for the image generator part. Hasn't the incredible extent of mode-collapse and the horrible DALL-E 3 style become universally know
468.
by gwern 2y ago
Original is https://www.endersanalysis.com/reports/blocking-free-world-t... but paywalled.
469.
by gwern 2y ago
Most Markdown dialects will support some sort of label or 'span' naming, which provides link targets (if only by dropping down to literal HTML <span>/<div> elements), and it looks like this does too: https:/
470.
by gwern 2y ago
> There aren't many personal like websites I remember that have stayed personal sites for over 10 years. But some have, kind of. Take gwern.net or stevepavlina (not a regular reader anymore): both of them - if you squint - look pret
471.
by gwern 2y ago
The content.lanl.gov server is hanging indefinitely and never serving the images. So it seems the images should be there on the server, the server is just dead in a weird way. I have emailed the journal and told them about the bug. In the m
472.
by gwern 2y ago
I used this to pick IBM Plex Mono for Gwern.net, which narrowly beat out the Adobe Source Code Pro which ranked highly & we had been using before.
473.
by gwern 2y ago
I'm also reminded of top-_k_ training/sampling, where you throw away the inconsistent samples: https://arxiv.org/abs/2002.06224 "Top-k Training of GANs: Improving GAN Performance by Throwing Away Bad Sam
474.
by gwern 2y ago
No, those points all fit perfectly. Birds don't look anything like a dog, so how does that make any sense to begin with? Why would the whole story's theme about dogs/humans suddenly introduce a third kind of animal? The narra
475.
by gwern 2y ago
I disagree with a few of the readings, too. Like the discussion of "air dogs" or "soaring dogs" in the linked translation: > Or again, the enigma of nourishment is easily solved when one understands that the dogs are
476.
by gwern 2y ago
I think what people are missing here is that this is for o1, and you are supplying questions & answers, but not the entire solution-solving transcript (as you almost never have such a thing). The whole point of o1 is that you don&#x2
477.
by gwern 2y ago
Yeah, I don't buy the theory either. Markdown has always made it easy to just drop in a bit of HTML wherever you need it, and major dialects like Pandoc Markdown support even more. You can set class, ID, and data-attributes on most ele
478.
by gwern 2y ago
I think that a lot of people feel like this post-COVID, like there's a sort of 'temporal collapse' where you feel stuck in a Big Now where Now never changes, no matter how many days do, and the pounds accumulate but not the y
479.
by gwern 2y ago
Yes, I've always read the story this way as well. Borges may have not been interested in politics, but politics was interested in him, and he clashed with the Peronists (who fired him from the library) and repeatedly criticizes fascism
480.
by gwern 2y ago
> I love thinking of this as such a wonderfully naive early 2000's view on HCI Not really so naive when you look into 'robotic process automation' https://en.wikipedia.org/wiki/Robotic_process_automati

More ›