justanotheratom

7 ms·

1. 61.▲
   by justanotheratom 1y ago

   Supabase's decision to take a dependency on Deno, IMO caused indirect pain to lot of devs. I have wasted quite a bit of time trying to find or load a package that I needed. And now, with Deno 2.0 apparently everything is node compatibl
2. 62.▲
   by justanotheratom 1y ago

   is there a well-established tool-chain for finetuning these models?
3. 63.▲
   by justanotheratom 1y ago

   FineTuning Ops platform requirement - compare evals, latency, cost across models.
4. 64.▲
   by justanotheratom 1y ago

   Is Best-of-N Sampling standard practice these days in Inference? Sounds expensive on the face of it. I am surprised because I thought the trend was towards cheaper inference.
5. 65.▲
   by justanotheratom 1y ago

   energy is abundant.
6. 66.▲
   by justanotheratom 1y ago

   I feel like these issues can be countered by a reasoning AI that runs locally and I can configure to operate in my best interest. e.g, Filter out political posts on X. Fact check opinion videos on the fly. I hope future computing devices wi
7. 67.▲
   by justanotheratom 1y ago

   Congratulations on the launch. Suggest adding the demo video right in the landing page. That is the first thing I would be looking for. IMO.
8. 68.▲
   by justanotheratom 1y ago

   Thanks for writing this. Suggest elaborating on the reason for using C++ in a Swift App.
9. 69.▲
   by justanotheratom 1y ago

   This is quite cool. Two questions: - why do you need nextjs frontend for what looks like a headless use case? - how much would be the OpenAI bill if there is 15 minutes of usage per day?
10. 70.▲
    by justanotheratom 1y ago

    Supabase is quite well integrated with vibe coding tools - cursor, replit, v0, etc. I agree that Firebase is a superior well-integrated product, but IMO the vibes are on Supabase side.
11. 71.▲
    by justanotheratom 1y ago

    I am looking for structured output at about 100-200 tokens/second on iPhone 14+. Any pointers?
12. 72.▲
    by justanotheratom 1y ago

    Anyone packaged one of these in an iPhone App? I am sure it is doable, but I am curious what tokens/sec is possible these days. I would love to ship "private" AI Apps if we can get reasonable tokens/sec.
13. 73.▲
    by justanotheratom 1y ago

    my thought exactly.
14. 74.▲
    by justanotheratom 1y ago

    just stating facts.
15. 75.▲
    by justanotheratom 1y ago

    Recently, OpenAI CFO Sarah Friar said, "We have something called A-SWE, which is the Agentic Software Engineer. And this is where it starts to get really interesting because it’s not just about augmenting a developer, like a Copilot mi
16. 76.▲
    by justanotheratom 1y ago

    are you predicting, or is there already a documented finding somewhere?
17. 77.▲
    by justanotheratom 1y ago

    Super cool. Imagine specialized hardware for running these.
18. 78.▲
    by justanotheratom 1y ago

    personas sounds great!
19. 79.▲
    by justanotheratom 1y ago

    bad example, sorry.
20. 80.▲
    by justanotheratom 1y ago

    This is an awesome direction. Few thoughts: It would be awesome if the custom rules were generalized on the fly from ongoing reviewer conversations. Imaging two devs quibble about line length in a PR, and in a future PR, the AI reminds abou
21. 81.▲
    by justanotheratom 1y ago

    exactly. I wish Groq would start selling their cards that they use internally.
22. 82.▲
    by justanotheratom 1y ago

    I thought English language was the Agent2Agent Protocol.
23. 83.▲
    by justanotheratom 1y ago

    neat, but instead of asking me to install browser extension, can you just bundle a browser in the MCP server?
24. 84.▲
    by justanotheratom 2y ago

    effects of excuses are thankfully limited to presidential term.
25. 85.▲
    by justanotheratom 2y ago

    F# was my favorite language, but - You have to chose the language of the Domain that you are working in, e.g, Swift for native iOS Development. Supabase Backend requires TypeScript, etc. - LLMs don't care about F#.
26. 86.▲
    by justanotheratom 2y ago

    Very interesting idea. Maybe you can add human in the loop ranking for CVs, which can be used to further tune the AI.
27. 87.▲
    by justanotheratom 2y ago

    That is a very crucial question. I am also interested in the answer. Perhaps they have RLS type policies that are only modifiable on the server.
28. 88.▲
    by justanotheratom 2y ago

    As someone trying to build a product solo, I agree that AI is making it easier to be a generalist. You do have to develop a sense of when to consult an expert vs relying on AI, or you might lose more time while saving money. Case in point -
29. 89.▲
    by justanotheratom 2y ago

    fixed the link. RE: How do either of those help or hurt Trump? I am not going to speculate that. But looking at the diff of the edit, anything of "substance" is removed from the response, and what remains is a non-answer.
30. 90.▲
    by justanotheratom 2y ago

    try using an LLM then.