nojs

12 ms·

181.
by nojs 1y ago
For json generation (and most API things) you should be using “structured outputs”
182.
by nojs 1y ago
This is how you win on LinkedIn: https://www.linkedin.com/in/patrickcollison > Dynamically communicate prospective opportunities and proactive technologies. Efficiently aggregate interactive materials before state o
183.
by nojs 1y ago
Why does he peel the oño like an orange when he’s about to cut it in half, why not cut it first?
184.
by nojs 1y ago
In other words, 2.7B -> 109M is a 96% reduction from headline to people. Could we apply the same maths to the 16B headline? I mean there’s not 16B people in the world, so a row per person can be ruled out pretty easily
185.
by nojs 1y ago
"Subagent orchestration" is also a really quick win in Claude. You can just say "spawn a subagent to do each task in X, give it Y context". This lets you a) run things in parallel if you want, but also b) keep the main a
186.
by nojs 1y ago
Right, it seems like these two use cases are rapidly diverging.
187.
by nojs 1y ago
> it's common for people to simply not have a clue how to even write the first stroke of a word they've heard and read before. It’s much more common to remember most of the character, or accidentally substitute it for a similar
188.
by nojs 1y ago
Even stronger than our need to anthropomorphize seems to be our innate desire to believe our species is special, and that “real intelligence” couldn’t ever be replicated. If you keep redefining real intelligence as the set of things machine
189.
by nojs 1y ago
To people who think this is surprising, it’s not that different to forgetting how to spell rare English words
190.
by nojs 1y ago
> I would really appreciate if someone has documented this extensively for docker compose Run `certbot certonly` on the host once to get the initial certs, and choose the option to run a temporary server rather than using nginx. Then in
191.
by nojs 1y ago
> For starters it works everywhere. Online. IRL. In my home country, in foreign lands. Haha, if only that were true. I’d say roughly 20% of my purchases are rejected because of a badly tuned fraud algorithm somewhere.
192.
by nojs 1y ago
I'm starting to think this is a deeper problem with LLMs that will be hard to solve with stylistic changes. If you ask it to never say "you're absolutely right" and always challenge, then it will dutifully obey, and alwa
193.
by nojs 1y ago
Could you explain your exact playwright setup in more detail? I’ve found that claude really struggles to end-to-end test complex features that require browser use. It gets stuck for several minutes trying to find the right button to click f
194.
by nojs 1y ago
Currently the quality seems to degrade long before the context limit is reached, as the context becomes “polluted”. Should we expect the higher limit to also increase the practical context size proportionally?
195.
by nojs 1y ago
In my experience jumping back like this is risky unless you explicitly tell it you made changes, otherwise they will get clobbered because it will update files based on the old context. Telling it to “re-read” xyz files before starting work
196.
by nojs 1y ago
How are you using GPT5 for coding (Cursor?)
197.
by nojs 1y ago
This type of “first shot success” is an easy test to run but not really indicative of how useful the tools are in practice. It’s more important to test how they navigate a large codebase, make reliable changes in line with existing code, no
198.
by nojs 1y ago
https://archive.is/w2aLn
199.
by nojs 1y ago
This pretty much matches my experience today. GPT5 (in Cursor) feels smarter in isolation, but CC with Opus is faster and better at real tasks involving a large codebase.
200.
by nojs 1y ago
As Homer would say, it isn’t ad supported so far !
201.
by nojs 1y ago
How does this ensure models haven’t seen it during training - is it a different benchmark per model release?
202.
by nojs 1y ago
So you can use an IDE other than VS code.
203.
by nojs 1y ago
Worth noting that a lot of people use those IDEs with vim keys
204.
by nojs 1y ago
The article implies that nobody pays usage-based pricing, but aren’t all API customers (ie essentially all business users) doing this now?
205.
by nojs 1y ago
Finally the Java logo makes sense.
206.
by nojs 1y ago
In my experience pretty much all of these issues stem from a combination of short context windows and suboptimal “context engineering”. If the agent has a clean, relevant context explaining what global functions are available it tends to us
207.
by nojs 1y ago
> Australia already requires you to register for voting and other things, so the trivial solution here is: give out anonymous time-limited tokens from the gov site Which “gov site”? Registering for voting does not give you an electronic
208.
YouTube to be included in Australia's teen social media ban(bbc.com)
16 points by nojs 1y ago|1 comments
209.
by nojs 1y ago
There is some argument to be made here about entropy, compression, and how if there are no surprises, the program communicates no new information. Interestingly, frequency of "surprising" sentences is one of the ways quality of AI
210.
by nojs 1y ago
Wow, that whole page does not inspire confidence. It’s 99% LLM slop. What We’re Doing: -We are making ongoing adjustments to our infrastructure to improve stability and ensure reliable scaling under elevated load -Analyzing system patterns

More ›