gsuuon

19 ms·

1. 301.▲
   by gsuuon 3y ago

   Didn't the human population drop to something like 2000 individuals at some point in history? Would be like dropping 2 rabbits in a box and opening it up to see 4 million.
2. 302.▲
   by gsuuon 3y ago

   There's also just far more tokens to train on if you do multi-language. I'd guess only the most popular languages would even have enough training data to get a specialized version - but it would still be an interesting trade off f
3. 303.▲
   by gsuuon 3y ago

   I'd really like to see smaller models trained on only one specific language, with it's own language specific tokenizer. I imagine the reduction in vocab size would translate to handling more context easier?
4. 304.▲
   by gsuuon 3y ago

   Y'all sure are some great students! ;)
5. 305.▲
   by gsuuon 3y ago

   You can also hide the logic away in F#: let calcDiscount = function | Percentage p -> subtotal * (decimal p / 100M) | FixedAmount f -> f And then: let totalDiscount = discount |> Option.map calcD
6. 306.▲
   by gsuuon 3y ago

   It is indeed assignment: https://learn.microsoft.com/en-us/dotnet/fsharp/language-ref... shadowing would be a second 'let a = 2'
7. 307.▲
   by gsuuon 3y ago

   > dotnet new list Shows all the available templates - you could give mvc or web a try?
8. 308.▲
   by gsuuon 3y ago

   I agree - I guess this article is more about a (very) gradual way to adopt F# into an existing C# code base, rather than an inherently useful combination of C# and F#.
9. 309.▲
   by gsuuon 3y ago

   I get not being bothered enough to opt out.. but actively opting in? What's the motivation there?
10. 310.▲
    by gsuuon 3y ago

    There is absolutely no benefit to me the user to force me to create an account. What a deeply frustrating user experience (especially when I have to dig around for my password just to update a graphics driver?!) I hope someone wakes up at N
11. 311.▲
    by gsuuon 3y ago

    Interestingly, mlc's web-llm runs much better on my iGPU than dGPU when the model size goes over available dGPU vram. Llama 2 7B runs faster on dGPU, but when I switch to Llama 2 13B suddenly my iGPU outperforms. I think because the iG
12. 312.▲
    by gsuuon 3y ago

    This is a neat one-liner if you don't want to install vim-fugitive on a temporary machine just for 'Gvdiff': set diff crb | vnew | exec "r !git show HEAD:#" | set diff crb Shows git diff with head - modify git show
13. 313.▲
    by gsuuon 3y ago

    I think this is the first time I've shed an actual tear for a famous person passing away.. RIP Bram Moolenaar. Vim was life changing for me and continues to be for many.
14. 314.▲
    by gsuuon 3y ago

    When I tried this maybe a year ago, I couldn't get actions/cache to work. It looks like they merged support - but I had to make other act-only changes to my workflow to accommodate it. I wonder if that's improved since then?
15. 315.▲
    by gsuuon 3y ago

    Yes this is more the tinkerer vs inventor approach, but sometimes tinkering is way more pragmatic in understanding a space.
16. 316.▲
    by gsuuon 3y ago

    This is incredibly close to what I currently do, except I journal by the day and some of the markers are different (I use `-,.=*[` to give meaning to each item) and items are organized hierarchically via indents. I used to use vimwiki, then
17. 317.▲
    by gsuuon 6y ago

    I had almost exactly that conversation recently. Now I wish I'd done my homework more, since that conversation led to a 1 year subscription. They've also already ran into the limits of the editor as a casual user on the premium pl
18. 318.▲
    by gsuuon 6y ago

    F# is slowly making it's way into C#... but you could also just skip the wait and start using F# :p