troupo

25 ms·

631.
by troupo 7mo ago
OpenAI is testing ads already: https://openai.com/index/testing-ads-in-chatgpt/
632.
by troupo 7mo ago
We were talking about this: https://news.ycombinator.com/item?id=47571715 and literally every single comment under this is talking about that. It's not a different topic. It's literally the topic of this branch
633.
by troupo 7mo ago
You want your training data to be clean or contaminated? A small number of samples can poison LLMs of any size https://www.anthropic.com/research/small-samples-poison --- start quote --- In a joint study with the UK
634.
by troupo 7mo ago
> So go back about one year, and we could vote about it before it got into the standard, and some of us voted no. Now we have a much harder problem. This is part of the standard proposal. Offtopic, but this is a problem in the web world
635.
by troupo 7mo ago
> You filter them, duh. Filters are also not 100% infallible > And you negotiate a contract where the seller bears some of that risk So the training data will be polluted anyway, but "the seller will bear some risk"
636.
by troupo 7mo ago
> Only if you assume that people who train models are stupid Someone in the chain will be. Even the smartest people buy a lot of their training datasets. What happens when those get contaminated?
637.
by troupo 7mo ago
Well, and then you have Claude Code which at one point needed 68GB of RAM to run https://x.com/jarredsumner/status/2026497606575398987 :)
638.
by troupo 7mo ago
> If the trees were in the same space as the panels, they'd be in the midddle of the parking space. What you'd have then is not a car park, but just a plain ordinary park. Sigh No, it's not. You can, and you should have
639.
by troupo 7mo ago
How is this better?
640.
by troupo 7mo ago
> I would go for a 2 or 3 hour walk with my phone using the remote control feature looking every 5 - 10 minutes 2-3 hours "walking" while having to check in every 5-10 minutes? If I have to check in every 5-10 minutes, I won&#x
641.
by troupo 7mo ago
> Claude should be shipped by a custom implementation of And when that fails for some reason it will happily write and execute a Python script bypassing all those custom tools
642.
by troupo 7mo ago
> having to answer for opinions with no basis in the literatu Having only literature on your side must feel nice. > They have access to the full transcript, and they have access to the full codebase, the diff history, whatever knowled
643.
by troupo 7mo ago
> Let's just include this for now to easily refute your argument which I don't know where it comes from: https://transformer-circuits.pub/2022/in-context-learning-an ... I love it when people include links
644.
by troupo 7mo ago
> So what do you think the difference is between humans and an agent in this respect? Humans learn. Agents regurgitate training data (and quality training data is increasingly hard to come by). Moreover, humans learn (somewhat) intangibl
645.
by troupo 7mo ago
> It does learn in context It quite literally doesn't. It also doesn't help that every new context is a new dawn with no knowledge if things past. > Also still not relevant, consider that agentic systems can be hierarchical
646.
by troupo 7mo ago
> because it has business context It doesn't because it doesn't learn. Every time you run it, it's a new dawn with no knowledge of your business or your business context > better reasoning It doesn't have better re
647.
by troupo 7mo ago
This is brilliant. Thank you!
648.
by troupo 7mo ago
Some of those "partners" may be storing your precise geolocation for 12 years: https://x.com/dmitriid/status/1817122117093056541
649.
by troupo 7mo ago
1. You don't have to use it all, but someone will. And there are over 200 keywords in the language: https://x.com/jacobtechtavern/status/1841251621004538183 2. On top of that many of the features in the lan
650.
by troupo 7mo ago
Lattner probably left because Apple didn't give the team any breathing room to properly implement the language. It was "we must have this feature yesterday". A lot of Swift is the equivalent of Javascrip's "we have
651.
by troupo 7mo ago
> @c @specialized @inline(always) @export(implementation) Reminds me of "In case you forgot, Swift has 217 keywords now" https://x.com/jacobtechtavern/status/1841251621004538183
652.
by troupo 7mo ago
This is still better than currebt monstrosities because the underlying select should still be an actual select (including keyboard navigation etc.)
653.
by troupo 7mo ago
> I don't know what that first one means. You mean the glass design? Not just glass. It started with Big Sur at least. It's forcing narrow and/or devoid of controls interfaces into every app, breaking decades-old system be
654.
by troupo 7mo ago
- They are rapidly iOS-ifying the desktop experience - All core services and apps experience significant performance degradation (to thenpoint that Spotlight regularly fails to find installed apps) which are currently only offset by the ins
655.
by troupo 7mo ago
> Macs are too expensive for the same performance/ram This hasn't been true for at least a decade. And it's especially not true for the M* series Macs. Even Macbook Neo can handle editing several layers of 4k video files
656.
by troupo 7mo ago
Macs are circling down the same drain.
657.
Is Argon2 better than Bcrypt?(pilcrowonpaper.com)
4 points by troupo 7mo ago|0 comments
658.
by troupo 7mo ago
This article is quite literally the only one that actually discusses actual Safari problems. And even this article falls prey to "failures in web platform tests" which are a very poor indicator. E.g. Safari passing all accessibili
659.
by troupo 7mo ago
> with iOS's ridiculous nonstandard interaction rules around when media is allowed to play. Are there any standard interaction rules on when media is allowed to play? I thought everyone implements it differently based on their own i
660.
by troupo 7mo ago
> I will say that all three(?) of them agreed to build it This is the key sentence > Chromium/Blink have a process, and it's quite rigorous It's Chrome's process, and Chrome's deadlines. > they have a r

More ›