lambdaba

31 ms·

1. 331.▲
   by lambdaba 3y ago

   More than that, it can monitor your screen continuously and have perfect recall, so it will be able to correct you immediately, or remind you of relevant context. I like to call it "Artificial Attention".
2. 332.▲
   by lambdaba 3y ago

   > Interruption Well, today is your lucky day!: https://persona-webapp-beta.vercel.app/ and the demo https://smarterchild.chat/
3. 333.▲
   Llama2.c running on a Silicon Graphics Indigo2 workstation(twitter.com)

   35 points by lambdaba 3y ago|10 comments
4. 334.▲
   by lambdaba 3y ago

   Thank you this is in line with what I was looking for!
5. 335.▲
   Ask HN: AI books worth buying in print?

   1 points by lambdaba 3y ago|2 comments
6. 336.▲
   by lambdaba 3y ago

   Is it as good as Perplexity though? I use ChatGPT for different purposes, I just thought that if Kagi would ally with Perplexity and benefit from its index (I'm not sure what Perplexity uses), it could get really good. I've only r
7. 337.▲
   by lambdaba 3y ago

   I'm only replying to the praise here - I too, although I haven't fully switched, had a very enticing moment with Kagi when it returned a result that couldn't even be found on Google at any page in the results. This really sol
8. 338.▲
   by lambdaba 3y ago

   Yeah, I guess I pressed tab right away and thought it was the design of the button. I think it's better to keep the focus style as the default outline, and uniform.
9. 339.▲
   by lambdaba 3y ago

   ah, yes, but since there's no focus indicator it's not obvious
10. 340.▲
    by lambdaba 3y ago

    If the author reads this, I think you shouldn't bind space to the side menu, I (hope) many people know you can scroll with space.
11. 341.▲
    Bespoke Automata, a GUI and deployment pipeline for making complex AI agents(github.com)

    2 points by lambdaba 3y ago|0 comments
12. 342.▲
    Bespoke Automata, a tool for building and deploying agent brains(github.com)

    1 points by lambdaba 3y ago|0 comments
13. 343.▲
    by lambdaba 3y ago

    I can't say exactly, but it degraded remarkably quickly. Maybe someone else can chime in with a different experience, but I doubt it would be anywhere near as good as what I read from MacBook users, especially M-series. I got the Think
14. 344.▲
    by lambdaba 3y ago

    That is remarkably low, isn't it? I mean, saying this as an (ex) ThinkPad user :P. Did you take any special care with it?
15. 345.▲
    by lambdaba 3y ago

    please share this, I'd like to check it out
16. 346.▲
    by lambdaba 3y ago

    Oh man. This was the Digg/Del.icio.us era. The Web was still the Web back then, decentralized.
17. 347.▲
    by lambdaba 3y ago

    I think one fair criticism is they could have had intrusion detection trigger when, presumably, the same IP address was logging in to thousands of accounts. But who knows how sophisticated the attack was? [edit]: there are other obvious heu
18. 348.▲
    by lambdaba 3y ago

    I agree with this. Frankly, if something is to blame, it's user agents / browsers that should be much better at educating and alerting about password reuse.
19. 349.▲
    by lambdaba 3y ago

    Yes, but you have to sit a little to find out...
20. 350.▲
    by lambdaba 3y ago

    AFAIK this is not something that's seen with psilocybin, maybe with frequent high dose LSD... Anyway, I wish people didn't even bring this up because it'd be a shame if that dissuaded anyone to try psilocybin.
21. 351.▲
    by lambdaba 3y ago

    Using gold nanoparticles.
22. 352.▲
    by lambdaba 3y ago

    I like the subtle styling changes in comment threads, less so in the submission lists.
23. 353.▲
    by lambdaba 3y ago

    BTW what do you mean by "different prompting style"?
24. 354.▲
    by lambdaba 3y ago

    Thanks, I've requested access, how long did it take for you? Update: access already granted!
25. 355.▲
    by lambdaba 3y ago

    I've recently subscribed too. The huge context window is a game-changer and I can't wait for API access. Unfortunately it hallucinates a lot. It's amazing for extracting information out of large documents, altough doing so ma
26. 356.▲
    by lambdaba 3y ago

    I regularly compare queries with both GPT-4 and Claude 2 and I find I prefer Claude's answers most often. The 100k token context is a major plus too. Claude is more creative and that comes with a higher rate of hallucinations. I hope I
27. 357.▲
    by lambdaba 3y ago

    Yeah, I couldn't say, it's just "vibes" I guess, just like the filler text produced by business consultants it's just not something that I feel would be missed if lost. It all looks the same at some level even when
28. 358.▲
    by lambdaba 3y ago

    To me it looks very analogous to AI-generated "art", it's very easy to generate some generally esthetically pleasing visuals, but the depth of the art stays in proportion with the input effort... Which is often not much. All
29. 359.▲
    by lambdaba 3y ago

    Well the "naked" reality is a commons, if you want an advertisement layer you put on your AR device.
30. 360.▲
    by lambdaba 3y ago

    Google Maps was really primitive at that time though :P.