antupis

13 ms·

1. 181.▲
   by antupis 3y ago

   I have gone oppositional direction younger me was kinda yes-man and now I am more jaded and see systemic failures all around me.
2. 182.▲
   by antupis 3y ago

   As old joke goes “There are two novels that can change a bookish fourteen-year old’s life: The Lord of the Rings and Atlas Shrugged. One is a childish fantasy that often engenders a lifelong obsession with its unbelievable heroes, leading t
3. 183.▲
   by antupis 3y ago

   also it might be enough now if they dominate some small niches at start and grow from there. Instagram started with mood pictures about coffee, tiktok was first about dancing videos etc. Zuckberg knows actual numbers and we don't.
4. 184.▲
   by antupis 3y ago

   I think biggest change will be how people make stuff so it will impact more in work than in personal space. Think something like if smartphone is car, LLM is more like truck.
5. 185.▲
   by antupis 3y ago

   Yup Haskell is those kinda languages that you always wanna learn but then hit bumb when you actually try to do something more complex than monad tutorial with it. I will definitely test this if it is something like Haskell--.
6. 186.▲
   by antupis 3y ago

   So instead of chunks of tokens, we can input stream of tokens and then some point say "LLM take a wheel". So it is very nice but not revolutionary.
7. 187.▲
   by antupis 3y ago

   Yup kinda best subreddit at the moment weird mix of llm training tips and tricks and people who wanna make sex bots.
8. 188.▲
   by antupis 3y ago

   It is rather weird that we are missing these kind workphones which would have good keyboard and good camera for zoom calls.
9. 189.▲
   by antupis 3y ago

   At least with Finnish GPT-3.5 and GPT-4 are only models that really work other are rather horrible.
10. 190.▲
    by antupis 3y ago

    I have visited several art exhibitions this year and generally would say that stuff at front page of /r/StableDiffusion/ is just more original and better than some random art exhibition.
11. 191.▲
    by antupis 3y ago

    Google 2023 is completely different company than it was in 2005. It is like comparing HP in 1999 and 1981.
12. 192.▲
    by antupis 3y ago

    I would say postgres is python of databases, second best to everything.
13. 193.▲
    by antupis 3y ago

    Yup, but the thing is that can you get model working how you want and can you run it production. My experience is that gpt4 yes, gpt3.5 yes but needs some work Llama2-70b 8-bit or more yes but needs some work after that it really depends us
14. 194.▲
    by antupis 3y ago

    It is little more complex than that there are especially elenentary school stuff which you need to learn doing mechanically(reading, simple Matt etc) and for those games migh be better than pen and paper. Lets put it this way if you need th
15. 195.▲
    by antupis 3y ago

    Adding features to a popular tool later can make it messy. Building a tool where you can add features easily makes it elegant eg Lisp.
16. 196.▲
    by antupis 3y ago

    I was thinking same, had hobby project where I did uploaded csv to DuckDb and then I use generate queries with chatgpt but building sematic layer top op duckdb sound much better.
17. 197.▲
    by antupis 3y ago

    M2 max studio is 2000$.
18. 198.▲
    by antupis 3y ago

    I think currently M2 max is best bang for buck running interface in open source model. But use case is so niche that Apple probably doesn't actively start supporting open source models. In the long run I hope some smaller company gets
19. 199.▲
    by antupis 3y ago

    There is lots of things like which you don’t want leak eg customer specific data. For those cases vectors are great.
20. 200.▲
    by antupis 3y ago

    I think some cases CF would be even good eg you want llm that produces only valid json data as output.
21. 201.▲
    by antupis 3y ago

    Yeah it is literally billion dollar question how to do that querying. Is there some good papers or blog posts that go through different approaches?
22. 202.▲
    by antupis 3y ago

    or junior could start doing more ops and less dev. In Finland, there appears to be a culture that values seniors way more and companies don't even hire juniors. This often results situations where seniors being saddled with time-consum
23. 203.▲
    by antupis 3y ago

    If they are very good examples then less is more https://arxiv.org/abs/2305.11206
24. 204.▲
    by antupis 3y ago

    yup and mainly for recruitment purposes.
25. 205.▲
    by antupis 3y ago

    it can be almost anything like your HN comments or some corporate wiki, then get colab pro 10$ month or some juicy gaming machine and fine-tune that using eg this tutorial https://www.philschmid.de/instruction-tune-llama-2
26. 206.▲
    by antupis 3y ago

    Call me paranoid but I would want bring my own device to defcon.
27. 207.▲
    by antupis 3y ago

    Somebody should make Tetris game where speed does not increase but dimensions do.
28. 208.▲
    by antupis 3y ago

    Well these models at least tell in name what they do eg llama-2-70b-Guanaco-QLoRA-fp16.
29. 209.▲
    by antupis 3y ago

    OpenAI would love that kinda regulation, it would basically kill free models.
30. 210.▲
    by antupis 3y ago

    Personally I hate it because it will snap you out from suspension of disbelief immediately.