zoogeny

34 ms·

481.
by zoogeny 3y ago
Have a look at the supabase-community claims repo [1]. NOTE: this is for ACL type permissions where you want to provide granular access to a wide range of services. This stuff is "really cool" but just keep in mind that it is pret
482.
by zoogeny 3y ago
One thing to keep in mind which I found using Postgrest interface: you will end up having to put logic into stored procedures. The rest APIs are actually very convenient for aggregating data like joins, but I started to get stuck as soon as
483.
by zoogeny 3y ago
I figured this must be a well-known property of neural networks. I'll do some reading on the lottery ticket hypothesis. That is almost exactly what I was thinking when reading the article: sure after you have trained it you can prune t
484.
by zoogeny 3y ago
I have a friend in real estate who uses matterport [1] to do something similar to this. They have tools to do animations very close to what is happening here. Even the circular cursor looks similar. 1. https://matterport.com&#x2F
485.
by zoogeny 3y ago
I'm not sure if that gives me an intuition on the title of the article: "Do large language models need all those layers" Am I interpreting you correctly if I say: "Finding the slope (training) may require those extra lay
486.
by zoogeny 3y ago
Forgive me if this is not a good question, but is there a difference here between training and inference?
487.
by zoogeny 3y ago
I'm a huge fan of Kaufman, I think I've seen almost everything he's done. In my opinion, "I'm Thinking of Ending Things" is not his best work. But I have to admit, I think of the themes in this movie more than
488.
by zoogeny 3y ago
I think it partially depends on what you are willing to forgive. For example, I am much more willing to forgive eagerness compared to carelessness or impatience. One thing to consider is that moving to a culture of "permissiveness&quot
489.
by zoogeny 3y ago
Your point about "worse jokes [...] on 4chan" is important. Wishing cancer onto someone is almost embarrassingly mild on 4chan. The idea that someone would take offence to that ancient insult is laughable. Outside of 4chan and wit
490.
by zoogeny 3y ago
> Everyone who memes long enough on the internet knows there's a meme about [...] As a counterpoint, I was working at a company and one of the guys made a joke in the vein of "I hope you get cancer". The majority of the pe
491.
by zoogeny 3y ago
First, I would need to know such a service exists. Given that I didn't know the service exists I would have to Google for such a service. Which is exactly the point in question. So, let's say I Google for such a service and I make
492.
by zoogeny 3y ago
Just an observation based on some people complaining that this isn't some significant advance over GPT-4 (even if it happens to actually be a small percentage gain over GPT-4 and not just gaming some benchmarks). One thing I consider i
493.
by zoogeny 3y ago
Wolfram Alpha has very much the same effect as Siri for me. Often I'll remember it exists and I will think to give it another try. But whenever I try it I get frustrated since it seems to fail on whatever I decided to attempt in that m
494.
by zoogeny 3y ago
Yes and no I think. I've seen individuals achieve things in their bedrooms that would make most corporations blush. Demoscene type stuff comes to mind as an example. Often a single person can become hyper obsessed with achieving some g
495.
by zoogeny 3y ago
You've got a lot of examples, but for example I recently thought: "How many weekdays are there between Jan. 11th 2023 and Mar. 11th, 2023" and got GPT to write the python code and run it to get the answer. It worked first try
496.
by zoogeny 3y ago
I like that this shows how hard even conceptually simple ideas are to achieve in fine-tuning LLMs. Even given a pretty good starting dataset, a decent starting model, etc. this appears to have been a challenge. One thing it did make me thin
497.
by zoogeny 3y ago
Tangentially, I am reminded of the essay "The Problem with Music" by Steve Albini [1]. There is a narrative in this essay about A&R scouts and how they make promises to bands. The whole point is that the A&R scout comes ac
498.
by zoogeny 3y ago
When people talk about existential risk fears, this is without a doubt my fear. All it would take is a small group that is ethnically supremacist and technically capable. How many edge-lords joke that the problem with the world is that ther
499.
by zoogeny 3y ago
This is the only response that actually addresses the article and my criticism. You make a good point that pressuring the governmental agencies that distribute funding to adopt a set of requirements could push the entire academic body in a
500.
by zoogeny 3y ago
That is fair, I have no reason to paint all academic publishing with a single broad brush. Personally, I am strongly on the side of free and open - probably in the direction of things like the Open Data Institute (ODI). But data, including
501.
by zoogeny 3y ago
I'm not an expert on ActivityPub nor academic publishing. What I'm trying to say is that what is built on top of the protocol has more to do with its suitability for a particular purpose than the foundational technology. The first
502.
by zoogeny 3y ago
This is a very interesting concept and one I think most comments here are missing. This isn't a question as to why Mastodon (or "Fediverse") has yet to catch on in the wider public. It is more a question as to why scholarly s
503.
by zoogeny 3y ago
I was watching a recent talk about the new Mojo language [1]. There is a section on SIMD and how they treat scalar values as a special case of vector operations (around the 33 min time). It does seem that tensors are one of the core abstrac
504.
by zoogeny 3y ago
> Friedrich Nietzsche [...] regarded morality as the ultimate form of passive aggression – a ruse employed by the weak, resentful masses to restrain the will of their stronger, creative superiors. I think this is an accurate but misleadi
505.
by zoogeny 3y ago
People opine that "we don't know what NNs are doing" and I think this is overblown while having some truth. But undeniably there are a host of unanswered questions arising from this latest surge in AI capabilities. It makes a
506.
by zoogeny 3y ago
To combat your pessimism - I would point to podcasts. In general, these were distributed on free platforms. The content became profitable through sponsorships. That is, the content creator does a deal directly with a brand to include a spot
507.
by zoogeny 3y ago
I have a golden retriever but he is a bit smaller than average. He isn't tiny, still around 65 lbs, which is the low end of the established breed size. If I bring him to the dog park he is about 85% the size of other retrievers. He was
508.
by zoogeny 3y ago
Interesting approach to cast the net so wide in a single release. I mean, I can imagine it makes a lot of sense for a company to just dump a bunch of documents into S3 and then expect an LLM to be able to answer questions on that corpus. In
509.
by zoogeny 3y ago
I do not own a DVD or Blu-ray player! I'm sure I could buy the DVD/Blu-ray on ebay (I saw it on Amazon for over $40!) and then buy a DVD/Blu-ray player. Or maybe it's on VHS or Laserdisc and I could obtain a physical cop
510.
by zoogeny 3y ago
Recently a friend of mine recommended me to watch the movie Heavenly Creatures. I went to justwatch.ca to see if it was streaming anywhere in Canada [1]. Not only was it not listed as streaming, it wasn't even listed to purchase anywhe

More ›