yeah what I'm trying to get across here is that: Dspy does not solve an immediate problem, which is why many feel this way and consequently why it doesn't have great adoption!
But on the other hand, I think people unintentionally end up re-implementing a lot of Dspy.
Source? We need the exact claim here, because there's a fine line between "we're in switzerland, so warrants aren't a thing!" (outright false) and "we're in switzerland, which have better privacy laws than other countries!" (debatable).
> Switzerland is a fundamentally different environment. Two of the things Switzerland is most famous for are also highly conducive to data protection: privacy and neutrality.
> When a law enforcement agency in the US requests user data from a Swiss company, it is illegal for that company to provide the data. At Proton, we reject all data requests from foreign agencies.
> Proton and other Swiss companies will only hand over user data when ordered to do so by a Swiss authority. And even then, Proton’s general policy is to challenge data requests whenever possible and only comply after all legal remedies have been exhausted.
So maybe your parent poster is confused? They do claim that being Swiss protects them from requests from foreign entities, but not Swiss entities. Which is what happened here, the Swiss authorities asked Proton for the data, then they handed it to the FBI.
Has Proton challenged the data and “only complied after all legal remedies have been exhausted”, though? That’s another question.
I wonder if the FBI knew it was going to be a pain in the ass asking for actual account access from the Swiss so they asked for financial records instead. Terrorism charges look pretty serious (regardless of how legitimate they are) so I'm sure that's what pushed the Swiss and Proton to comply.
But the Swiss have the notion of a warrant, no? So if a Swiss judge or official issues a proper warrant, then a Swiss company or citizen is obliged to comply with it.
Read the wording, they aren't complying with US warrants, they are complying with Swiss-issued warrants. US LE/Intel agencies figured out this loophole some 20 years ago.
"classical ML" models typically have a more narrow range of applicability. in my mind the value of ollama is that you can easily download and swap-out different models with the same API. many of the models will be roughly interchangeable with tradeoffs you can compute.
if you're working on a fraud problem an open-source fraud model will probably be useless (if it even could exist). and if you own the entire training to inference pipeline i'm not sure what this offers? i guess you can easily swap the backends? maybe for ensembling?
not really. I've downloaded balatro. I saw that it was moddable. I wrote a mod API to interact programmatically. I was just curious if, from text only game state representation, a LLM would be able to make some decent play. the benchmark was a late pivoting.
why not use a classification of food that actually aligns with what is bad? it seems like we don't actually know. Nova combines a bunch of different attributes some of which we don't actually think are causally linked to bad health.
People do this, and the good ones don't have anything to do with processed food, or if they do, it's entirely superfluous. The Mayo Clinic publishes on this topic and, as I recall, strongly recommends the Mediterranean diet - high in fiber and protein, nutritionally diverse, inclusive of fats and carbohydrates.
I know multiple people that are drinking litres of olive oil daily because of the Mediterranean diet. Because of this critical oversight, I am forced to conclude it's bogus. A real recommendation would address this.
I don't understand anything you're saying. A diet can not compel you to do anything, let alone drink liters of olive oil. I assume this is attempting to parody something about Nova but I frankly can't unpack whatever it's supposed to be.
reply