I told you a true story about my experience and you call it 'hyperbole'. Well, ok if it suits you.
On your second point, proper developers release security updates and backports for software released decades ago, while others always chase new shiny things and contribute a lot to this unsupportable legacy hell that we have to deal with.
I'm done w/ hybrid cameras. Just make a video focused camera or a photography focused camera. Didn't people learn their lesson after seeing how well FX3 did?
We have examples today of companies that don't need to justify an eventual PE ratio of 25 so why would this be a thing all the sudden? I think you all are misunderstanding vague notions of market fundamentals as some kind of rules by which the market works.
You are in fact literally required to purchase an Apple or Google phone to participate in an OECD economy. Your employer will require it, your government will require it. Increasingly, your banks and doctors will require it. Unless you're telling people to live off the grid? But maybe we can have something between the two extremes of living in a fucking jungle or submitting to an enforced duopoly wherein two companies control what billions of people can publish and take a 30% cut from it all? No? That too unreasonable for you?
Rightfully so. Your entire premise is that design should be made with accessibility in mind whereas in reality, accessibility tools should be built to handle those concerns and design should be free to do what it needs to do full stop. If that's being accessible that's fine, but trying to enforce accessibility standards is like saying that Levis needs to make pants for people without legs.
Well it seems like it’s over for litellm. You’d have to be negligent to be using them at this point. If you’re managing this at your company good luck explaining why you should continue to use litellm after this
I appreciate the feedback. We work hard to make every release safe and secure. The attack was novel, and I'd love your feedback on how we can improve our ci/cd further.
"Novel" is doing a lot of work here. Save the PR for people that will fall for it. There's nothing novel about being internally compromised which you've demonstrated two times in a row
Alignment is a very broad term. True, it encompasses things like censorship and mandatory refusals in areas people will never agree 100% on, but a completely-unaligned LLM would be next to useless. When the system prompt tells the model, "You are a helpful assistant," that's alignment.
Well if we are going to be broad enough about "alignment" to just mean training that makes it practically useful, then yeah, it's a flawed premise and people are wasting their breath talking about it at all. It's basically broad enough that it means nothing
Seriously. It feels like Anthropic is realizing this too with their new attempt at safeguards. Trying to train the model to "only do good things" was not working, and so they opted to just bolt a topic classifier on the outputted text.
My suspicion is that model makers are going to give up on what people currently mean when they say 'alignment' and instead rely on additional processing of the model output so they can have their cake (a veneer of safety) and eat it too (latent space reasoning, etc.)
Well that explains a lot about the doomerism and misinformation we regularly find on HN. If you don’t see the agenda being pushed or look into who is funding the documentary for just about every documentary in existence then you’re just not very intellectually honest. For example one of the vegan pushing ones on Netflix is straight up funded by/closely tied to for-profit artificial meat companies
However, it's great that developers are not too concerned with backwards compatibility. It was really holding back a lot of software
reply