IanCal

67 ms·

631.
by IanCal 1y ago
> of the 14 stories I see when I visit the page, 4 of them (more than 25%!) Llm maths? ;)
632.
by IanCal 1y ago
If you’re working with them I’d like to highlight that if they have a messaging platform with children on they are going to have to take safety extremely seriously. I know the laws in the uk are not popular here but the checklists of risk a
633.
by IanCal 1y ago
> Babies on airplanes get a pass because iykyk I think broadly if you’re doing something you just need to do, then a kid being a kid (particularly babies) is fine. Even if it’s annoying, that’s just life. Beyond that you need to pick you
634.
by IanCal 1y ago
This is a code editing model. 10 tokens per second editing may as well not exist for any interactive use case.
635.
by IanCal 1y ago
> are they going to more important to the economy than humans?", then they have to be good at basically everything a human can do, I really don’t think that’s the case. A robot that can stack shelves faster than a human is more valu
636.
by IanCal 1y ago
Why do you think any of these should be a challenge for, say, O3/O3 pro? You pretty much just have to ask and give them access for these things. Talking to a stakeholder and translating jargon and domain terms? Trivial. They can churn
637.
by IanCal 1y ago
No, vLLM is a thing for serving language models: https://github.com/vllm-project/vllm
638.
by IanCal 1y ago
This is atrocious. > There exists a class of questions in life that appear remarkably simple in structure and yet contain infinite complexity in their resolution space. Consider the familiar or even archetypal inquiry: "Darling, ple
639.
by IanCal 1y ago
I'd just like to highlight that this is one of quite a few things of a similar incredible amount of detail that they have done: https://ciechanow.ski/archives/
640.
by IanCal 1y ago
> I think the ability to swap out APIs just isn't the bottleneck.. like ever It's a massive pain in the arse for testing though. Checking which out of X number of things performs the best for your use case is quite annoying if
641.
by IanCal 1y ago
Do we have the vocab? That's part of the point here. Does it take images? How are they tokenised?
642.
by IanCal 1y ago
A really important thing is the distinction between performance and utility . Performance can improve linearly and utility can be massively jumpy. For some people/tasks performance can have improved but it'll have been "in
643.
by IanCal 1y ago
Are you sure you're using the same models? G2.5P updated almost exactly a week ago.
644.
by IanCal 1y ago
Ish - it always depends how deep in the weeds you need to get. Tokenisation impacts performance, both speed and results, so details can be important.
645.
by IanCal 1y ago
>If entire commit is generated by AI then it is obvious what created it - it’s AI. Whether it's committed or not is irrelevant to the conclusion there, the question is what was the input .
646.
by IanCal 1y ago
> So it can save calendar events if you type something into a prompt They didn't type it. They hit a button, it suggested adding it to the calendar and they approved it. You can still find some of the flow forced, but I think you&#x
647.
by IanCal 1y ago
> OpenAI CEO Sam Altman has voiced worries about the use of AI to influence elections, but he says the threat will go away once “everyone gets used to it.” Then he's lying or a complete moron. People have been able to fake things fo
648.
by IanCal 1y ago
> - something that can be stored for the cost of less than a day of this industry's work Far, far less even. You can grab a 1TB external SSD from a good name for less than a days work at minimum wage in the UK. I keep getting surpri
649.
by IanCal 1y ago
Good find - the first video is a frequency sweep, video 2 has some music. Edit - I'm not sure that's the same thing? The release talks about pixel based sound, the linked paper is about sticking an array of piezoelectric speakers
650.
by IanCal 1y ago
This is where I find having a disconnect between an ML team and product team is so broken. Same for SE to be fair. Accuracy rates, F1, anything, they're all just rough guides. The company cares about making money and some errors are mu
651.
by IanCal 1y ago
So, I'll start by saying I love that you have made a thing and shipped. That's great, and the following is intended to be helpful. I read this and understand what it is. I then went to the site and it says bathroom privacy, tells
652.
by IanCal 1y ago
Never use anything other than the absolute bleeding edge immediately after release? That's frankly wild, and totally ignores that many people release new versions while supporting older ones for transition.
653.
by IanCal 1y ago
> but how often do you really want to explore the docs in a specific version? Absolutely any time I'm not using the latest version?
654.
by IanCal 1y ago
> and then to explicitly submit a request to that LLM that explicitly said to read and parse arbitrary issues (including the one created earlier) and then just blindly parse and process and perform whatever those issues said to do, and t
655.
by IanCal 1y ago
I assume you posted this as a counter, but it really isn't. It doesn't distinguish between wet and dry amd (the claim is very specifically for dAMD) but more importantly something that can help prevent a problem is very differen
656.
by IanCal 1y ago
That overview I can find it talking about wet AMD, the claim for this is specifically dry AMD. edit - and from the paper > Notably, while ROCK inhibitors have been previously suggested for treatment of wet AMD and other retinal disea
657.
by IanCal 1y ago
I'm not sure I really follow, you can create new tables for any step if you want to do it entirely within the db, but you can also just run duckdb against your dataframes in memory.
658.
by IanCal 1y ago
What's the issue you have? The idea is you have random testing and the test failures are added as explicit tests that then always get run. Is that so different from someone else testing? The main issue is you stumble across a new issue
659.
by IanCal 1y ago
This is fair, but also a tooling or process problem. The co-worker should add the test to a new branch / test it on main. If it fails that's a new ticket (with the great side effect of having a failing test). If that passes it&#x2
660.
by IanCal 1y ago
I used it for running a series of "execute this api call". If it's valid for a user to do, you can make a list and have it do those in sequence. I had this for a UI library. It could call the functions to add and create the l

More ›