For the less knowledgeable on generative AI amongst us (👋), what is VFM?
I came up with Visual Foundation Model, am I on the right track?
Love the examples btw, agree it's a nice benchmark!
I hope so too, and if that's the case that other groups can learn how to replicate it!
(note: I'm not at all suggesting Scheibenbogen's group is doing anything wrong, just wondering whether there may be some sort of hidden structural bias to recruitment or the study itself)
From my own very brief "literature review" (I mostly just scanned key sections of a few papers) I got the impression that most or all of the studies in which patients have seen improvements through IA were from Scheibenbogen's group.
Concerned that improvements are due to some kind of bias
If it wasn't happening through widespread laundering of licensed code through a provenance-obscuring laundry machine — sure!
Is there any logic in classifying anything as FSD? Isn't it just basically saying that they don't know what is wrong?
I see IBS is often listed as FSD, but IBS patients indisputably often respond to eliminating certain foods, which would suggest that the pathology is just not known
It does apply to all LLMs realistically available for the task today though.
Sorry, I misread your original post and didn't see your second one.
At least to me, when it comes to coding, the lack of consent for training material is basically THE ethical complaint.
An LLM trained on material for which consent is given is ethically all right for this purpose, imo.
truly novel, but you can't know. The point is that the LLM essentially laundries the licensing by obscuring the provenance of the code that it spits out.
I think LLMs can be very useful, but I do personally think it's ethically problematic to use them for many aspects of coding
This might be hard to prove, but we're debating morality here, not whether you're able to prosecute successfully.
With LLMs there is no such "step by step". It might give you a verbatim copy of some code, it might give you something that's essentially a derivative of licensed code, or something ...
For example, if you, a human dev, were to copy some licensed code, and then make some superficial changes to it so that it's no longer a verbatim copy, then your code would still generally be considered derivative and subject to original licensing.
Not the person you're responding to, but a human dev knows whether or not they're essentially just copying code. An LLM doesn't know and doesn't tell you, yet it's able to reproduce non-trivial licensed code more or less verbatim. You just wouldn't know.
Det er så sinnsykt at leger kan si sånt til tross for alle de negative erfaringene pasienter rapporterer om. Hva skjedde med føre-var eller "først, gjør ingen skade"?
Oppriktig lei meg for at det skjedde med deg!
Uff, det var leit å høre. Er SIPCOV det prosjektet ved Kysthospitalet som de publiserte denne artikkelen på?
jamanetwork.com/journals/jam...
That would be very surprising given that Natelson has worked with ME/CFS for a long time, and was on the committee that developed the IOM criteria...
But I also couldn't find any info on recruitment. I found something that may or may not be the study protocol. Better wait for the paper, I guess.
Kanskje det bare var nevnt i ulike sammenhenger, uten at det ble gjennomført systematiske studier på det?
Takk, det var interessant å høre på!
Jeg er litt kjent med både Liens og Systrom sitt arbeid, og så vidt jeg vet har de kun gjennomført 2-dagers CPET, ikke over tre dager slik Beate nevnte. Det hadde vært interessant å se resultater fra en slik studie, men jeg har ikke klart å finne noen.
If, like me, you wake up every morning wondering what the Voronoï diagram of arbitrary shapes looks like, here is the answer:
Oh come on, you can't say things like that without showing plots! 😂
Very well said (all of it).
Scientific progress is so intertwined with our historical and cultural development, I am concerned what might happen if humans are no longer a substantial part of the process.
Garner, not Gardner. Well, close enough.
I'm not trying to downplay health anxiety or mental distress brought on by post-viral disease, these are very real issues.
But pretending that Paul Gardner's story is representative of LC patients as a whole is so incredibly disingenuous.
Wow. There are just several layers of sad irony here. For one, she responds to your criticism of her cherry-picking with... A single anecdote. I almost laughed!
And for that anecdote she chose Paul Gardner, who by his description seems to have suffered more from health anxiety than anything else.
It's strange to me to see such a massive study, which must have cost a lot of money, on ... Paxlovid? It never seemed like a likely good candidate to me. It seems as if a much smaller study could have yielded much the same insight at a fraction of the cost.
From the paper:
«Recently, the Association for the Treatment of Neuroplastic Symptoms considers PPS/MUS to be ‘neuroplastic’ [Citation3,Citation4].»
So because they don't understand, it must be psychosomatic? It's like doing inverted science: start with the conclusion, don't bother with evidence
I can't find any stated reasons for dropping out. That makes it very hard to reason about... Unfortunate that they don't seem to have asked and included the responses in the data.
SSRIs can have some crazy side effects, which could also be partly to blame for higher drop-out rates
Ich glaube aber, dass der Begriff bei manchen Krankheiten und in manchen Kontexten anders interpretiert wird.
In mehreren Artikeln wird „exercise intolerance“ mithilfe von CPET gemessen, was eher für deine Interpretation spricht.
Die Definition im DocCheck Flexikon (auf Deutsch) macht jedoch deutlicher, dass der Begriff auch Reaktionen auf die Aktivität umfasst:
flexikon.doccheck.com/de/Belastung...
Jetzt bin ich mir auch unsicher!
Der englische Wikipedia-Artikel sagt, dass der Begriff auch Schmerzen, Fatigue usw. nach der Aktivität umfasst, aber dafür gibt es keine Quelle 🤔 Ich habe keine eindeutige wissenschaftliche Definition gefunden.
Err, Zeichenlimit!
