lukev

40 ms·

631.
by lukev 3y ago
I am extremely glad to see RDF being worked on and improved. It's an extremely powerful, general data model. It gets a bad rap because if its unfortunate association with the XML syntax and the fact that many of the reference implement
632.
by lukev 3y ago
This is a question I've always had about the debt industry... How hard would it be for consumers (or through a proxy) to seek out and buy their own debt? You mention in another comment that medical debt (for example) could go for pen
633.
by lukev 4y ago
Given that this seems to describe a real problem, why hasn't anyone launched a proxy service that normalizes across the varying implementations? Why should tens of thousands of developers have to discover the particular quirks of an im
634.
by lukev 4y ago
But embedding-based semantic search can handle arbitrary sized databases. I fully believe context windows are going to grow: I am skeptical they will grow to cover "all your company's documents" or even "the full encyclo
635.
by lukev 4y ago
The search+prompt approach has another benefit, which is that allows a chat interface to "cite it's source", something you really can't do with just parameters (fine-tuned or not.) Although a lot of people are working on
636.
by lukev 4y ago
It shouldn't be made illegal. The problem is that while the workforce is the most important part of the whole enterprise, each individual worker has very little bargaining power. Workers should be able to form some sort of group, so th
637.
by lukev 4y ago
Has anyone done any comprehensive analysis on exactly how much quantization affects the quality of model output? I haven't seen any more than people running it and being impressed (or not) by a few sample outputs. I would be very curio
638.
by lukev 4y ago
I agree that LLMs could be one module in a future AGI system. I disagree that LLMs are good at simulation. They're good at prediction . They can only simulate to the degree that the thing they're simulating is present in their tr
639.
by lukev 4y ago
You could do those things in theory. I'm not saying that you could never build AGI on top of a LLM, or that such a AGI could not become "misaligned." I'm just saying that having a mental state that's natively in Eng
640.
by lukev 4y ago
We apparently have misaligned understandings of what we mean by "how they work." I agree, we don't know how to interpret the weight structure that the model learns during training. But we do know exactly what happens mechanic
641.
by lukev 4y ago
On one hand I hate to belabor this point, but on the other I think it's actually super important. Both of things things are true: 1. The relationships between parameter weights are mysterious, non-evident, and we don't know precis
642.
by lukev 4y ago
Whether it has inner state is highly relevant to my claim, which was that the only state a LLM has (aside from its parameters) is transparent and readable in English. Which the context is.
643.
by lukev 4y ago
Responding here since it won't let me continue your thread any more. No, there's a fundamental misunderstanding here. I'm not saying the model will tell you the truth about its internal state if you ask it (it absolutely will
644.
by lukev 4y ago
I don't need examples. It's simply how they work. This is why they hallucinate. A LLM is fundamentally a mathematical function (albeit a very complex one, with billions of terms (a.k.a parameters or weights)). The function does
645.
by lukev 4y ago
This is exactly the kind of mysticism I'm talking about. In fact we know precisely how LLMs work. The fact that parts of human linguistic concept-space can be encoded in a high dimensional space of floating point numbers, and that a
646.
by lukev 4y ago
I agree with this 100%, while also disagreeing with the "robot god will kill us all" objections to AI which unfortunately tend to get most of the mindshare. I think it's important to realize that these are two _completely sep
647.
by lukev 4y ago
The reasoning pattern of Yud and his ilk is fundamentally theological and eschatological in nature. There isn't any actual understanding of the technology involved. It is fundamentally an ontological argument. Because they can imagine
648.
by lukev 4y ago
Seems similar to https://github.com/rqlite/rqlite or https://dqlite.io . Would be interesting to see a breakdown of the differences.
649.
by lukev 4y ago
Unlike information embedded in the parameters, a LLM has the capability to "cite its source" for information in the context window.
650.
by lukev 4y ago
If the argument is that it's very easy to emotionally confuse language models and persons, than I reject that argument on the following grounds: No works of fiction are persons. All "I" statements from the current generation
651.
by lukev 4y ago
There is a ton in this article and it's very thought provoking, you should read it. But I think it ignores one critical dimension, that of fictionality . There is plenty of text that people would ascribe 'personhood' to acco
652.
by lukev 4y ago
I'm not sure it counts as "evidence" because I can't show it to you, but based on personal introspection, whenever I write something it starts with a barely-linguistic thought I want to communicate that gradually gets ex
653.
by lukev 4y ago
I think this is a real concern, but imagine a couple other scenarios: 1. You have a widely read spouse named Joe who reads constantly. He's got a good memory, and typically if you have a question you just ask him instead of searching f
654.
by lukev 4y ago
Oh, I agree. Optimistically, the existence of LLMs will be a forcing function for education to focus on what actually matters. (IMO, this is the ability to critically assess and generate arguments , something GPT isn't particularly
655.
by lukev 4y ago
I agree with some of this (particularly the conclusion that better education methods are required) but lets be a bit generous for a second. The ability to write well is (or was) an important skill. Being able to use correct grammar, to stru
656.
by lukev 4y ago
One thing I have a hard time getting away from as a consumer is the delta in graphics quality between VR and a high-res display. Modern AAA games on a 32" 4k monitor that I'm sitting right in front of are pretty damn beautiful and
657.
by lukev 4y ago
We do have "socialized law care" for criminal cases in the USA. That's what a public defender is. If you cannot afford a lawyer one will be provided by the court. That is a constitutional right. Of course they are overworked,
658.
by lukev 4y ago
That's because it's a large language model, not an AI. Anyone calling it an AI is buying/selling the hype cycle, or (at best) using an over-general term for ML in general.
659.
by lukev 4y ago
Yes, the article covers those (some of them don't fall into the time period under discussion.) There's a few possible drivers that are no doubt helping but it does appear that between one thing and another, chess itself is going v
660.
by lukev 4y ago
I was going to post that I suspected this was due to some sort of automated or bot activity... a couple months seems like way too short a time for that much organic growth. However, Google Trends also indicates a doubling of interest in che

More ›