cjbprime

30 ms·

451.
by cjbprime 4y ago
Not just claws, but feet! I also like the ouroborous nature of the snake becoming the scroll, or perhaps being created by its own writing. (No, I don't know why.)
452.
by cjbprime 4y ago
> To date, no neural network shows a capacity for reasoning and creativity worthy of an animal level. Current LLMs are autoregressive models that predict the probability distribution of the next token given the set of previous tokens cal
453.
by cjbprime 4y ago
https://commoncrawl.org/
454.
by cjbprime 4y ago
She explains in the article that she's a free-to-play player and has not spent any money on it.
455.
by cjbprime 4y ago
> These fine-tuned models are intended for research use only and are released under a noncommercial CC BY-NC-SA 4.0 license, in-line with Stanford’s Alpaca license. This is a no-commercial-use-allowed license; it is neither considered fr
456.
by cjbprime 4y ago
GPT-4 is actually multi-modal, not text. ChatGPT does not yet expose image upsubmission to it. But it's part of how the model was trained already.
457.
by cjbprime 4y ago
> it can only know what we teach it since it relies on all of our perceptions to learn > I think it’s emergent behavior has many of the hallmarks of reasoning and novel inference You contradicted yourself here, right?
458.
by cjbprime 4y ago
To be clear, this is not "what does this line of code do?", but "what does this many-line program do?". Even if you've seen all the lines individually before, you still have to work out what happens when you combin
459.
by cjbprime 4y ago
How is it completing the task? The only possible way to perform that task reliably is to have a principled understanding of what the code actually does, and to create a world model based on that understanding, and then extract consequences
460.
by cjbprime 4y ago
> Imagine a person taking a test. This person has the power to look at everyone else's answers No, that analogy fails. ChatGPT can pass new tests that it has not seen before.
461.
by cjbprime 4y ago
> of course machines can think and be alive. Among my colleagues, who are mostly senior-and-above software and security engineers, I seem to be a strong outlier in saying "yes, that bot can obviously perform reasoning". I&#x
462.
by cjbprime 4y ago
What a disappointing article. So many reasonable explanations of technically correct descriptions of forward passes, backprop, gradient descent, and then at the end: > but ChatGPT has no actual conception of what it’s talking about—no i
463.
by cjbprime 4y ago
> And of course, they have zero understanding of anything at all. They’re a next-token-predicting machine with an unimaginably large training set. Oof, this fallacy is so frustrating. Being a next-token-predicting machine doesn't im
464.
by cjbprime 4y ago
I do think GPT-4 has emergent reasoning skills, but I don't think this particular example is very complicated, and you must always be careful about whether you're describing a situation it will have learned all of the relevant det
465.
by cjbprime 4y ago
I guess I'm with the LLM on this one, since I can't follow your example. Did the sign flip over while it was falling? Did the sign fall towards or away from the glass door that I am on the other side of? Where are the doorhandl
466.
by cjbprime 4y ago
It clearly knows what words are. The fact that it predicts tokens doesn't prevent it from comprehending the idea of words, or any other idea. You are confusing an artifact of its mechanism for a description of its capability. > It &
467.
by cjbprime 4y ago
I think maybe they just want to handle small API requests for many millions of devices at once.
468.
by cjbprime 4y ago
> Year over year investment in AI decreased, after many years of consistent increase. Well, that seems blatantly untrue. Maybe they stopped collecting data for the report in, like, mid-2022?
469.
by cjbprime 4y ago
The rest of the global economy depends on encryption; cryptocurrency is a tiny portion of that. (In that context, worrying about cryptocurrencies after a core cryptographic algorithm is made totally vulnerable feels like worrying about whet
470.
by cjbprime 4y ago
I suspect that rather than AI cannabilizing VR/AR interest, it will accelerate it, as VR/AR provide compelling personas to powerful chatbots.
471.
by cjbprime 4y ago
I read that as more "we cold emailed people looking for a potential contact" than "we submitted this vulnerability to their PSIRT". The fact that they say this is not a vulnerability disclosure situation suggests that th
472.
by cjbprime 4y ago
> As an aside, I’m really starting to hate these threads on here. I'm not sure what to say, other than that if you'd like to have less frustrating conversations, you could do better than showing up with hearsay where someone as
473.
by cjbprime 4y ago
I can't tell whether they attempted to disclose it to Toyota through normal vulnerability disclosure channels, though. The article implies to me that they didn't.
474.
by cjbprime 4y ago
Great article. But: > And part of the problem is that this isn’t a vulnerability disclosure and so the processes that Toyota does have in place are not appropriate. I didn't follow this part. I hear that the authors think their &qu
475.
by cjbprime 4y ago
I don't think you're being clear about whether the questions were novel. If you discover your question was uncreative, surely e.g. some details, wording, facts, names, or numbers inside the question can be changed to defeat a mod
476.
by cjbprime 4y ago
Is ARC a benchmark that GPT-4 can be tested against today? I would be curious to see its results.
477.
by cjbprime 4y ago
If it's a stochastic parrot, then merely randomizing proper nouns and filler text should be enough to prevent its abstraction ability. If you're saying that we can't use a problem if any analog of that problem has ever been d
478.
by cjbprime 4y ago
> something that I know is only working from words as a source. Your intuition about the architecture might be causing the distrust. Yes, the ends of the model are words in and words out (although GPT-4 is now multi-modal and can accept
479.
by cjbprime 4y ago
New bar for people claiming LLMs can't reason: invent a specific, testable problem, representable in text, that many humans can solve and LLMs can't, and tell us what it is.
480.
by cjbprime 4y ago
This is one of those comments that begs for actually using LLMs. Yes, an advanced LLM like GPT-4 will absolutely show evidence of an abstracted world model, correctly answering questions about where person X thinks object Y is, or what rela

More ›