afro88

22 ms·

331.
by afro88 2y ago
When I try it it just says "Not found"
332.
by afro88 2y ago
I think the advice is based on "peace time" management, during more comfortable profitable times, leaning towards more centralised control. There's no appetite for the risk of moving faster by removing approval bottlenecks or
333.
by afro88 2y ago
>> Decentralize decision-making authority as much as possible. Remove barriers in their way, slash approval layers, attack dependencies. > This is awful advice. You can only operate in this mode for at best 2-3 months before your e
334.
by afro88 2y ago
The o1 model is hit and miss for me. On one hand it has solved the NYT Connections game [0] each day I've tried it [1]. Other models, including Claude Sonnet 3.5 cannot. But on the other hand it misses important detail and hallucinates
335.
by afro88 2y ago
The solution of the cipher example problem also strongly hints at this: "there are three r's in strawberry"
336.
by afro88 2y ago
Seeing the "hmmm", "perfect!" etc. one can easily imagine the kind of training data that humans created for this. Being told to literally speak their mind as they work out complex problems.
337.
by afro88 2y ago
Why would it be most likely the former?
338.
by afro88 2y ago
I'm skeptical that this would be easy to learn. I haven't thought about what row certain keys are on for about 25 years. I imagine there would be a long period of frustrating typing and not being able to get things done as fast as
339.
by afro88 2y ago
Can you elaborate on how to do that without duplicating style configurations?
340.
by afro88 2y ago
> often-wrong The author was very careful with their words: they didn't say the calculator was wrong. They said it was confusing and sensitive to small adjustments. It's likely that the same confusion and variable sensitivity e
341.
by afro88 2y ago
> Often this takes the form of trying to access an endpoint that does not exist, or read a property off the API response that does not exist. After probing a bit more, my suspicions are usually confirmed — ChatGPT hallucinated that endpo
342.
by afro88 2y ago
Move away from the songs strawman you're arguing against. My point was about content in general. As we get better and better at recommending the next content item based on engagement and likes, we make things more addictive. We don&#x2
343.
by afro88 2y ago
No, but that's also not what I'm saying. Open your mind a little bit. Here's my reply to someone else: https://news.ycombinator.com/item?id=41303877 There is a world where services are engaging, fulfilling an
344.
by afro88 2y ago
Those are non-starters, but there is a lot that can be done in the space. Not autoplaying the next item is a good example. Look at how services handle things in kids mode. That's just one way to think about it.
345.
by afro88 2y ago
> All documents were public by default. Technical and product RFCs (and later PR/FAQs) were drafted, reviewed, and catalogued in a public Google Drive folder Does this still exist somewhere?
346.
by afro88 2y ago
You missed my point a little. We keep optimizing for more addictive services. It's just as dystopian for a service to always present exactly what will keep me using the service and giving thumbs up.
347.
by afro88 2y ago
Yeah, me too. But this being google, and them talking about youtube... you know where this tech is going next.
348.
by afro88 2y ago
> Using transformers to incorporate different user actions based on the current user context helps steer music recommendations directly towards the user’s current need What if the user's current need is to not play music? To not con
349.
by afro88 2y ago
Github may have it wrong then? Their summary says: > Permissions of this strongest copyleft license are conditioned on making available complete source code of licensed works and modifications, which include larger works using a licensed
350.
by afro88 2y ago
Is it? Doesn't it depend on whether your stack depends on it to function, or if it is standalone that your stack can connect to but doesn't rely on?
351.
by afro88 2y ago
"Because I like solitude and find space and weightlessness fascinating" is also a fine response (my response to why I would also find 8 months in space something I could embrace). The "why" of the GP was a personal quest
352.
by afro88 2y ago
That's incredibly interesting, thanks for sharing. I'm not aphantasic but when deliberately picturing something the image starts correctly but changes to something else quickly. They are also not vivid at all. However when I think
353.
by afro88 2y ago
What am I looking at, and how is it related to SAM2?
354.
by afro88 2y ago
Banks seem fine now. At least ING and NAB.
355.
by afro88 2y ago
If it helps, the videos are of kids being rescued mere seconds after showing drowning signs. Watching these gives you very quick training on what those signs are, in a way that images or instructional pamphlets can't.
356.
by afro88 2y ago
Have a chat to the developers and see if having copilot / chatgpt available has influenced how they break their PRs down first.
357.
by afro88 2y ago
It genuinely took me 5 seconds to realise the first image was a photo from inside the Sphere, as opposed to the article being about spherical spaces in caves.
358.
by afro88 2y ago
> We use a host of quantitative and qualitative measures. None of them show any positive improvements. These include the basics like roadmap reviews, demo sessions, feature cycle time, etc as well as fairly comprehensive business metrics
359.
by afro88 2y ago
Writing code is just one part of the process. Other bottlenecks might prevent you from seeing overall productivity improvements. For example: - time between PRs being created and being picked up for review and merged - time spent on releasi
360.
by afro88 2y ago
Fine tuning doesn't quite work that way. You have to format the training data set as request/response. The idea of fine tuning is to get the model to output things in a specific format, style or structure. Your use case is better

More ›