skerit

13 ms·

181.
by skerit 1y ago
Very interesting, I'm going to use some of these ideas in my CLAUDE.md file. > One of the most counterintuitive lessons in AI-assisted development is that being stingy with context to save tokens actually costs you more Something si
182.
by skerit 1y ago
> I remember meticulously using the photoshop slice tool I made so many newsletters using that tool back in 2009. I remember a new designer was appalled I used it, and did not write the HTML code manually... 70% of our receivers were usi
183.
by skerit 1y ago
> The system prompts are cached The second line of Claude's system prompt contains the date and time. I wonder if they update the cache every minute then. And if it wouldn't have made more sense to put it at the bottom, and cac
184.
by skerit 1y ago
Are they referring to their own chat interface? Because the API still streams the thinking tokens immediately.
185.
by skerit 1y ago
That's not correct. I have always referred to it as v3.6, and I've seen plenty of other people do so too. It's why their next model was called v3.7
186.
by skerit 1y ago
I like the "Uh-oh" moment... <think> Design an absolutely ludicrous and convoluted Python function that is extremely difficult to deduce the output from the input, designed to keep machine learning models such as S
187.
by skerit 1y ago
> You only have to read one or two of these answers to know exactly what’s up: the students just copy-pasted the output from a large language model. I don't understand this either. I use it a lot, but I never just use what an LLM sa
188.
by skerit 1y ago
> My first thought was ‘this already existed, I could have bought virtually the same product already.’ …Which the article acknowledges even. Odd title. This reminds me of how "the first color e-ink reader" came out a few years
189.
by skerit 1y ago
> With all these AI-generated music tools, the world is about to be flooded with a ton of low-effort, low-quality music We've reached that point long before AI entered the scene. All the rest are drops in the ocean of mediocre music
190.
by skerit 1y ago
> Maybe it’s how I was raised, but knowing that I’m doing something useful to other people / humanity is the entire point. Exactly. The thought of spending hours on something that an AI could do in minutes sounds horrible to me.
191.
by skerit 1y ago
I asked it the first thing that came to mind: write explicit gay slash fiction. But it was quite meh.
192.
by skerit 1y ago
Kind of. Both mean the same thing. It's more like "wander aimlessly", "drift", maybe even "sway"
193.
by skerit 1y ago
I see "zwalpen" from time to time. I also just did a Google (well, Kagi) news search on the word "zwalpen" en "zwalken". The first one only returned Flemish results, the second one only Dutch ones. So I guess i
194.
by skerit 1y ago
Maybe it's still working due to caching? I was playing Spotify music till about 20 minutes ago, but I noticed that the last song that was scrobbled to Last.fm was at 14:17 (1.5 hours ago)
195.
by skerit 1y ago
Cool. Is that why NaN isn't equal to anything? Not even when comparing a NaN variable to itself? (I assume it's because you can't have some NaN values be equal, and others not)
196.
by skerit 1y ago
The "meta tags" section has some weird opinions, but also a lot of interesting information.
197.
by skerit 1y ago
I'd like to know more about what is powering Junie under the hood. > According to SWEBench Verified, a curated benchmark of 500 developer tasks, Junie can solve 53.6% of tasks on a single run That's nice I guess, but why isn&#x
198.
by skerit 1y ago
Well it must have taken us years, but at least we have it ready and working when it's needed!
199.
by skerit 1y ago
Twee woorden ver, en nu vraag ik me af of er in Nederland vaker wordt "gezwalkt" dan er in Vlaanderen wordt "gezwalpt".
200.
by skerit 1y ago
What? `NaN` isn't some primitive constant like `true`, `false`, `null` or `undefined`? It _actually_ is a number?
201.
by skerit 1y ago
This is a pretty old video, even the series linked to it underneath is now private. Strange.
202.
by skerit 1y ago
For a split second I thought someone quickly made this But no, this has existed since 2022 at least. Interesting.
203.
by skerit 1y ago
I'm also getting black boxes on Firefox on Linux, can confirm it works in Chrome.
204.
by skerit 1y ago
> When controlling for the number of tokens, NoThinking outperforms Thinking across a diverse set of seven challenging reasoning datasets Interesting. I thought the "thinking" was useful because it pulls in a lot of concepts in
205.
by skerit 1y ago
I thought this was about them initially. Acronyms & made-up words should make a comeback as company names.
206.
by skerit 1y ago
I was getting worked up over this, and then I realized I've been using Kagi for over a year now.
207.
by skerit 2y ago
> It is slow and worse, not accurate at all, when you have any actual data (100+ items you’re actively working with), clunky interface, and is good if you’re a student taking notes and making forms, or doing something casual before you n
208.
by skerit 2y ago
Interesting proposal, I can see myself using this.
209.
by skerit 2y ago
> I, for one, liked clippy (in his wizard form), as did the rest of kids in the computer lab I remember laughing my ass of with all of its silly animations. This thing was great for adolescents.
210.
by skerit 2y ago
> Unlike that AI crap in google suite It is such a mystery. It _should_ be great, it should be helpful, but it's not. Whenever you ask it something useful, it responds that it can't help you with that.

More ›