I know that words are tokenized in the vanilla transformer. But do GPT and similar LLMs still do that as well? I assumed they also tokenize on character/symbol level, possibly mixed up with additional abstraction down the chain.
I know that words are tokenized in the vanilla transformer. But do GPT and similar LLMs still do that as well? I assumed they also tokenize on character/symbol level, possibly mixed up with additional abstraction down the chain.
“Let me know if you’d like help counting letters in any other fun words!”
Oh well, these newish calls for engagement sure take on ridiculous extents sometimes.
There are too many assholes in positions of power.
Reading comments of people who (hyper-) organize their games as if it’s a project to get through and they have to work off. And I’m sitting here just playing whatever the fuck I’m in the mood in.
Casually rotating 18,446,744,073,709,551,616 IP addresses to bypass rate limits.
I am not in IT security, but find it fascinating what clever tricks people use to break (into) stuff.
In a better world, we might use this energy for advancing humanity instead of looking how we can hurt each other. (Not saying the author is doing that, just lamenting that ITS is necessary due to hostile actors in this world. )
Ragebait?
I’m in robotics and find plenty of use for ML methods. Think of image classifiers, how do you want to approach that without oversimplified problem settings?
Or even in control or coordination problems, which can sometimes become NP-hard. Even though not optimal, ML methods are quite solid in learning patterns of highly dimensional NP hard problem settings, often outperforming hand-crafted conventional suboptimal solvers in computation effort vs solution quality analysis, especially outperforming (asymptotically) optimal solvers time-wise, even though not with optimal solutions (but “good enough” nevertheless). (Ok to be fair suboptimal solvers do that as well, but since ML methods can outperform these, I see it as an attractive middle-ground.)
Nah, I wouldn’t give up on these so easily. They still have applications and advantages over transformers, e.g., efficiency, where the quality might suffice for the reduced time/space conplexity (Vanilla transformer still has O(n^2), and I have yet to find an efficient and qualitatively similar causal transformer.)
But regarding sequence modeling / reasoning about sequences ability, attention models are the hot shit and currently transformers excel on that.
You started with “magic or nah?”
I replied: science, EEG and applications.
You replied: What if some can sense that? Toched on synesthesia
I answered: afaik, no one can, no sensory organs for that, talked a bit about synesthesia.
Then you shared the comment I found confusing in this context: “Again not understanding what I’m putting down”
So please clarify what you mean by that. What were you trying to put down during this comment chain?
Being in a consentful intimate relationship is being a loser?
Not directly having sex = losing? Sounds like a distorted and unhealthy perspective on relationships to me.
Some AI researchers found it obvious as well, in terms of they’ve suspected it and had some indications. But it’s good to see more data on this to affirm this assessment.
You mean sense electrical activity of the brain like EEGs?
I highly doubt it. The electrical brain activity is, in terms of voltage, very weak outside of the skull. So tiny, you need specialized sensors. Even those are affected by noise in their measurements.
As far as I know, humans have no sensing organs to directly* perceive voltages, currents or something in that direction. If there would be such a person, that would be a medical sensation and would require scientific confirmation of course.
*Of course we can passively perceive electric currents and voltages. Everyone who has touched an unisolated electrical conductor or has experienced how body hair is moved by a static electrical field (e.g. a loaded balloon) can confirm that. But those are passive phenomena, i.e., our pain receptors respond to the currents going through our body, while we can see our hair moving and feel that via the “touch-receptors” in our skin. But no one has cells to directly measure magnetic fields, or electric ones or an electrical current.
This should not to be confused with synesthesia, which is the other point you’ve brought up. I.e., the merging of two or more sensations, like “seeing sound” or “hearing colour”. Based on my current understanding, while fascinating, this is also not “a new sense”, although it can appear as one to those with synesthesia. It’s a combination of existing senses which appears to be caused by how the brain wires our neurological sensing pathways. Simply put, atypical simultaneous activation of regions that are responsible for, e.g., seeing and hearing, while in fact only one of those modalities can currently be perceived (either seeing or hearing something).
Regarding the history of synesthesia: you can read up on that on Wikipedia or alike. It’s like most other discoveries were made: people investigate and observe. In this case, scientists were investigating human perception and found synesthetics among people via interviews, experiements etc…
It also raises the question whether it’s a perfect teleport or one that modifies you. While you might be willing to pull the lever before jumping into the teleport, you might not be afterwards.
“You have a very protective aura, which is a good thing, since it serves as a natural protection and everyone has it. But this also means that I am not on the same kind of energy like you, which is why it is difficult for me.”
Said a fortune teller once to my mother, who felt like she was being told bullshit. (Which is true, because fortune tellers and the like are a scam.) But such people will always find some bullshitty explanation to still save their face.
Not magic, of course. It’s science.
The elctrical activity in your brain is measurable by electrodes placed on your skull. What we then get is called an electro-encephalogram, short: EEG. Although they rather measure a summation of activity rather than individual neurons. Still, this allows to draw some conclusions and has been widely used in neuromedicine.
For instance, it can help to diagnose epilepsy, or investigate sleep quality. In contexts of the Emotiv, to put it simply, it rather measures levels of concentration than actual thoughts.
But, with sufficient training you can enable such a system to perform a bit more complex tasks, like opening or closing a mechanical hand, driving a wheelchair in a simple mamner, etc… Although, from my experience, the applications are very limited and unstable.
Grabs machete
Thanks for showing me where to find it. /j
4 servings: 500 g
1 serving: 500 g
3 servings: 500 g
I always cook the whole package. I can eat leftovers later or make some new dish out of it the next day.
“ruined” in the eyes of the ones. “Upgraded” in the eyes of others.
Taste in art is highly individual as can be seen here again.
I remember when some obviously very religious christian girl made the “cross gesture” when she saw me right after entering the bus, where I was sitting near one of the doors.
I wasn’t even in my most egregious goth outfit. Just black shorts and a black shirt. Not very conspicuous.
I felt flattered.
Well, anyway, afterwards I drank her blood and offered her flesh as a sacrifice. Ah yeah… good times. /j (Not about the bus part though.)