lxe

24 ms·

1. 331.▲
   by lxe 3y ago

   So basically "Internal tools that are fun to build make bad startup ideas" If you build a cool dashboard, that might not sell as well as some sort of ugly integration with some horrible legacy system.
2. 332.▲
   by lxe 3y ago

   Always learning something new from Josh W Comeau
3. 333.▲
   by lxe 3y ago

   If ccTLD is under control of groups like Taliban, and more instances like this surface, how likely it is for ICANN "revolt" and intervene in ccTLD operations and take over?
4. 334.▲
   by lxe 3y ago

   Nice to see Reboot is back!
5. 335.▲
   by lxe 3y ago

   We had radio buttons forever, but no, let's complicate things.
6. 336.▲
   by lxe 3y ago

   I've been listening to Radiolab for a few years now on my runs, and I generally enjoy it a ton. Occasionally they really really dumb things down to the point of making me wonder who they think their audience are. But oftentimes they pr
7. 337.▲
   by lxe 3y ago

   Love seeing demoscene on HN. Excellent work.
8. 338.▲
   by lxe 3y ago

   https://rammb-slider.cira.colostate.edu/ has a UI for this
9. 339.▲
   by lxe 3y ago

   I use the CSU's RAMMB SLIDER UI at https://rammb-slider.cira.colostate.edu/
10. 340.▲
    by lxe 3y ago

    I wonder what the system prompts for this is, and what's the underlying model? One can use this to generate a lot of synthetic conversations and then use counterfactual / inverted DPO to finetune an actually useful model.
11. 341.▲
    by lxe 3y ago

    This is just Anthropic's new Claude release.
12. 342.▲
    by lxe 3y ago

    Do what exactly? I have no issues with GPU split on oobabooga with either exl2 or gguf.
13. 343.▲
    by lxe 3y ago

    Does ollama support loaders other than llamacpp? I'm using oobabooga with exllama2 to run exl2 quants on a dual NVIDIA gpu, and nothing else seems to beat performance of it.
14. 344.▲
    by lxe 3y ago

    I'm in love with this idea but the interface is just so clunky. Clustering needs a bit of massaging and zooming in and out should be smoother. Also labels and clicking on places without opening a new page would be great.
15. 345.▲
    by lxe 3y ago

    I've used baofeng as a base for a while. I put a home-made yagi on the roof of a shed and I could communicate with probably a few dozen active repeaters, some hundreds of miles away. Had to rotate the yagi via a state of the art "
16. 346.▲
    by lxe 3y ago

    They should "un-bin" the statistics and identify individual outliers -- those who scored poorly on the tests but hold higher GPA, and vice versa, and learn things from these individual students.
17. 347.▲
    by lxe 3y ago

    My first home automation circa 2006 was ejecting CD tray pushing a garage opener button. The server hosted a WAP page accessible via my pre-smartphone cellphone to close and open the garage.
18. 348.▲
    by lxe 3y ago

    It's an ollama interface, not an actual model implementation in Ruby. Title should be "INTERFACING with Open-Source AI Models Locally with Ruby"
19. 349.▲
    by lxe 3y ago

    I'm also struggling with it despite having it on my bookshelf for a while. Not an easy read. It's cited and referred to all over the place, so maybe it's worth getting through it.
20. 350.▲
    by lxe 3y ago

    Gotta get a bigger antenna for my baofeng to receive this.
21. 351.▲
    by lxe 3y ago

    Notch is Markus Presson, the creator of Minecraft. He got $2.5 billion in cash from Microsoft when he sold it. In the years since he had a whole lot of controversial tweets, but that didn't' seem to affect him due to his wealth.
22. 352.▲
    by lxe 3y ago

    This is more of a classic lesson in what effect can alcohol have on one's life and career.
23. 353.▲
    by lxe 3y ago

    This is very cool! Gonna give this a shot.
24. 354.▲
    by lxe 3y ago

    Oh this is neat! I was wondering how to get whisper to stream-transcribe well. I have a similar project using whisper + styletts with the similar goal to gave minimal delay: https://github.com/lxe/llm-companion
25. 355.▲
    by lxe 3y ago

    Would love to see the process of this pasta being made.
26. 356.▲
    by lxe 3y ago

    Yes
27. 357.▲
    by lxe 3y ago

    Here's the chat log if you're curious: https://chat.openai.com/share/6386ad1f-a44d-4cc1-a5d1-0e3c3c... Sega Enterprises Ltd. v. Accolade, Inc.: In this 1992 case, the court held that Accolade's reverse e
28. 358.▲
    by lxe 3y ago

    I was researching whether there's legal precedent making such uses legal, and ChatGPT pointed out that "Sony Corporation of America v. Universal City Studios, Inc.," where it was decided that developing a VCR where users can
29. 359.▲
    by lxe 3y ago

    I think anything compatible with either chat completions or completions API should work.
30. 360.▲
    by lxe 3y ago

    Apparently he's been in prison for murdering his wife since 2008