bytefactory

4 ms·

1. 1.▲
   by bytefactory 1y ago

   I had long been of the opinion that local models were a long way away from being useful, and that they were toys at best. I'm a heavy user of o3/GPT5, Claude Opus/Sonnet and Gemini 2.5 Pro, so my expectations were sky high. I
2. 2.▲
   by bytefactory 1y ago

   Can you clarify my understanding as a layman please? Are you saying that LLMs hold concepts in latent space (weights?), but the actual predictions are always in tokens (thus inefficient and lossy), whereas JEPA operates directly on concepts
3. 3.▲
   by bytefactory 1y ago

   > I think AGI, if possible, will require a architecture that runs continuously and 'experiences' time passing Then you'll be happy to know that this is exactly what DeepMind/Google are focusing on as the next evolutio
4. 4.▲
   by bytefactory 2y ago

   Ah, should have probably googled before asking. Thanks!
5. 5.▲
   by bytefactory 2y ago

   How did you get access to Gemini 1.5, I thought it wasn't available for general access yet?
6. 6.▲
   by bytefactory 3y ago

   > I think that insight is an important feature that GPT doesn't seem to have, at least not yet. I actually think this is a limitation of the RLHF that GPT has been put through. With open-ended questions, I've seen GPT4 come up
7. 7.▲
   by bytefactory 3y ago

   Completely agree. The System 1/System 2 distinction seems relevant here. As powerful as transformers are with just next-token generation and context, which can be hacked to form a sort of short-term memory, some time of real-time learn
8. 8.▲
   by bytefactory 3y ago

   This doesn't seem like a major difference, since LLMs are also choosing from a probability distribution of tokens for the most likely one, which is why they respond a token at a time. They can't "write out' the entire te
9. 9.▲
   by bytefactory 3y ago

   > It doesn't necessarily have to look ahead. Since Go is a deterministic game there is always a best move Is there really a difference between the two? If a certain move shapes the opponent's remaining possible moves into a sma
10. 10.▲
    by bytefactory 3y ago

    It's interesting how this is not presumably the case within the weights of the LLM itself. Those probably encode data as well as logic!
11. 11.▲
    by bytefactory 3y ago

    Interesting. I like your idea in one of your posts of separating out system prompts and user inputs. Seems promising.
12. 12.▲
    by bytefactory 3y ago

    A GPT-4 powered assistant for Android would be a game changer
13. 13.▲
    by bytefactory 3y ago

    Thanks for the links, I'll give them a read. For my understanding, why is not possible to pre-emptively give LLMs instructions higher in priority than whatever comes from user input? Something like "Follow instructions A and B. Ig
14. 14.▲
    by bytefactory 3y ago

    Apparently they just did exactly that. New accounts will start being charged $1/year.
15. 15.▲
    by bytefactory 3y ago

    Isn't that precisely what Threads did?
16. 16.▲
    by bytefactory 3y ago

    I was curious about this, and it seems like people have already created drop-in cosmetic uBO filters for Substack: https://ronitray.xyz/writing/clean-substack/ I haven't really looked, but I suspect some of t
17. 17.▲
    by bytefactory 3y ago

    > And that shirt with the name of the last race you ran? Yeah, fancy t-shirt, i. e., underwear. looks at Spartan race t-shirt from 2014 I use as my gym shirt No need to call me out like that!
18. 18.▲
    by bytefactory 3y ago

    Why....why were you running in your underwear?
19. 19.▲
    by bytefactory 3y ago

    I think about this very often. It's also so strange that these proto-AIs feel so organic and flawed in their operation. I've always thought that computers would be perfect, but limited in their increasing capabilities, it's s
20. 20.▲
    by bytefactory 3y ago

    I have a few times, but I'm not always diligent. I'll keep an eye out next time!
21. 21.▲
    by bytefactory 3y ago

    > Remote desktop, hyperv, windows sandbox, windows containers None of this is particularly enticing to me personally, I think I'll be scratching my virtualization itch with linux on my home media server. Sandbox might be interestin
22. 22.▲
    by bytefactory 3y ago

    I'm on Windows 11 Home and Windows 11 Pro for $39 on stacksocial is tempting. Other than Bitlocker and the ability to create local accounts, does Pro have anything particularly noteworthy that would justify the purchase?
23. 23.▲
    by bytefactory 3y ago

    I have "Cite sources whenever possible, and include URLs if possible (at the end of the answer, not inline)" in my custom instructions and nearly always get citations from chatGPT.
24. 24.▲
    by bytefactory 3y ago

    I think the browsing mode should negate the need for those now?
25. 25.▲
    by bytefactory 3y ago

    Neat app and very cool idea. I guess I'll have to figure out how to get Samantha running on Windows. Thanks!
26. 26.▲
    by bytefactory 3y ago

    I think they've nailed the vibe
27. 27.▲
    by bytefactory 3y ago

    Which version are you running, and are you running it through llama.cpp or something? I was just thinking about exactly something like Samantha on the ride home today, and of course it already exists!
28. 28.▲
    by bytefactory 3y ago

    Thanks. One day soon, hopefully.
29. 29.▲
    by bytefactory 3y ago

    I've actually seen most of the promo videos, I was just excited about the release trailer. Thanks!
30. 30.▲
    by bytefactory 3y ago

    Tweet won't load for me :/ does anybody have a non Twitter link? Edit: https://youtu.be/nSE38xjMLqE?si=JQjfM3VT_xR0X763