I love Nike shoes and brand but they just don’t produce anything since fly knits that I actually like. All I want is a flat shoe with a massive toe box and thin material for the gym with Nike branding - why is it so difficult to make something people want?
Not sure, the labs will probably just cripple the security features of these models for a while I think and even potentially put back doors into systems for the security services…
I've never found feature flag systems complex and you can build one yourself in a few functions. The services I have used in the past were extremely simple to integrate so I don't understand the issue either way honestly.
The referendum was done specifically in two parts - 1. do you want us to try to find a deal with the EU 2. We will thrash out any details, get the best deal we can with the EU and give you another vote.
Seems like Iceland didn't want to even waste the government's time looking into this... I quite like their government treated everyone like adults.
Most people do not enjoy the output of the open models or know where their code is being sent so I’m out. Also I don’t trust Elon anyway, I’m just going to make a VS Code Extension that handles things how I want but I’d like to offer the same deals Cursor got for tokens with OpenAI and Anthropic, for me this was the key 1/4 token cost!
Why you would trust Elon Musk to not hack you and steal your model weights is beyond me, but maybe they intend to use that compute just for training, still I’m sure there’s plenty of research there to take too…
No, you were deliberately disingenuous (I understood that) but I decided to answer the best possible interpretation of what you meant. I'd seriously reconsider this argument style, just because it's common doesn't mean any of us should think it's acceptable to pass of half truths as pithy commentary. You won't listen of course, nobody does anymore, they just want some ego points of being technically right in the narrowest of ways. Pointless point scoring.
Yes, they could also sell me GPT Sol 5.6 or 5.7 on a chip and I’d probably buy it. It’s a really really useful model for me, I’m not sure how much better for coding I need it to be. For most things I find Sol good enough with a small amount of coaxing around my tastes.
I’m willing to bet we can see longevity improvements from better blood sugar control. Not 100% certain but extremely likely, this would include potential heart attack and dementia reduction…
Yes, if it cannot be measured it doesn’t exist no matter how much it would make everyone feel more comfortable if we could intuitively understand these systems.
The original RoPE paper is a great example of the authors not actually understanding how their own discovery was functioning fully which is a fun example of what I’m talking about I guess…
Yes you might progress the field, but will you really understand why? You can make up an explanation and anthropomorphise it with a few contrived diagrams and everyone will cheer!
I think we are talking about different things to be honest. Understanding LLMs is a fine thing to do but to say you understand what the network is actually doing, even in toy examples like MNIST is extremely tricky. Figuring out what a ~3 trillion parameter network is doing in thousands of dimensions seems intractable to me.
Linking to a Twitter thread about training a small LLM to be good at Wordle is not an example of what I'm talking about. It might well be a useful task but it doesn't allow us to understand deeply what's happening.
What provable conclusions have you intuited around how LLMs work? Give me some examples to prove your point? I'm absolutely happy to change my view with enough data.
I'm not sure it's possible to have intuition about systems that work in thousands of orthogonal dimensions. In fact I'm pretty sure most of the research is people trying fairly arbitrary things and testing them and then post rationalising implied understanding of what is really happening on top of good outcomes.
I bet you 100% it was built not by in house team but farmed out to an agency for £££££ millions. They will have been saying in their case studies some absolute nonsense about how they saved the government millions rather than burned loads of money too.
I was thinking how tricky it would be to learn to program now, honestly a sense of laziness is one of the reasons I learned programming - so that I could automate away repetitive tasks. Our generation might be the last people who can actually program properly, everyone will just learn some basic architecture and point the AI in vaguely the right direction.
Can the US even afford a trade war, they are beginning to find borrowing more money increasingly difficult… currently the debt is around $130k for every man, woman and child in the US.