yboutros

joined 1 year ago
[–] [email protected] 1 points 2 days ago

Hilarious, any bets on how much not changing those credentials cost them

[–] [email protected] 1 points 3 weeks ago

I'm sure both are good, I've just heard Xen is better for isolation. I'll check out spectrum-os, although I am starting to get used to the NixOS monolithic kernel architecture, an linux OS with nixpkgs is the next best thing

 

Meaning, VMs with Xen and hardware virtualization support

The system VM/Qube for USBs is isolated, the Network VM/Qube is separate and isolated, the windowing system and OS housing the qubes is isolated....

And being able to configure all of those with Nix would be a wet dream come true

[–] [email protected] 1 points 1 month ago
[–] [email protected] 1 points 1 month ago

Thanks for the feedback! I also asked a similar question on the ai stack exchange thread and got some helpful feedback there

It was a great project for brushing up on seq2seq modeling, but I decided to shelve it since someone released a polished website doing the same thing.

The idea was the vocabulary of music composition are chords and the sentences / paragraphs that are measures are sequences of chords or sequences of measures

I think it's a great project because the limited vocab size and max sequence length are much shorter than what is typical for transformers applied to LLM tasks like digesting novels for example. So for consumer grade harder (12GB VRam) it's feasible to train a couple different model architectures in tandem

Additionally, nothing sounds bad in music composition, it's up to the musician to find a creative way to make it sound good. So even if the model is poorly trained, so long as it doesn't output EOS immediately after BOS, and the sequences are unique enough, it's pretty hard to find something that isn't different that still works.

It's also fairly easy to gather data from a site like iRealPro

The repo is still disorganized, but if you're curious the main script is scrape.py

https://github.com/Yanall-Boutros/pyRealFakeProducer

[–] [email protected] 3 points 1 month ago

Underrated comment

Everyone's conspiring folks. What's hard to measure, is who's conspiring

[–] [email protected] 2 points 1 month ago (1 children)

Laplacian Edge detection? beautiful meme

[–] [email protected] 1 points 1 month ago

I'm not crazy about ads, it can be nice when they're effective for the right reasons. That said, a break in happened at my estate once and the only evidence I needed or could secure was because I left my phone at home that day

[–] [email protected] 28 points 2 months ago (8 children)

I wish more guys just said they didn't know something instead of clearly not knowing what they're talking about and running their mouth based on vibes

[–] [email protected] 0 points 2 months ago* (last edited 2 months ago)

I sort of agree, but I think it depends on effort.

Type one word in and try and sell the easiest generated image? Low value.

But typing the right combo to create assets to create something larger than the model is capable of? That's more valuable.

Criticizing AI or artists that leverage AI is like criticizing an artist for using a printer instead of drawing by hand

Or saying someone's digital work is inferior because they used a tool to help make their image...

On that note, when working on a large project, is an AI artist as pretentious as the artist in the comic because they got some help generating the project from an AI instead of another human? Or is someone's work ethic less credible for Google searching instead of asking a person? Are works of art valuable because they're entirely original and uninfluenced by anything else but the artist themself? Because with that metric no artists are valuable since nothing is entirely original anyways

[–] [email protected] 23 points 2 months ago (2 children)

25% of reddit comments are chatgpt trash if not worse. It used to be an excellent Open Source Intelligence tool but now it's just a bunch of fake supportive and/or politically biased bots

I will miss reddits extremely niche communities, but I believe Lemmy has reached the inflection point to eventually reach the same level of niche communities

[–] [email protected] 6 points 2 months ago (2 children)

Don't tell him, if too many people get ad blockers they're just going to keep evolving

 

When training a transformer on positionally encoded embeddings, should the tgt output embeddings also be positionally encoded? If so, wouldn't the predicted/decoded embeddings also be positionally encoded?

[–] [email protected] 10 points 3 months ago (1 children)

You're right, we need water fountains with milk instead

 

Went through the pain of packaging a python project on Nixos. Here's some issues I hit, and how I got lucky resolving them. I feel the most reliable way of doing this in the future is to use docker and just imperatively build.

Here's how I got web drivers, AI dependencies, gpu dependencies, and an api dependency bundled together into an ephemeral shell for python development, on NixOS 23.11

  1. Enable Flakes

  2. Start with setting up poetry2nix

  3. Get the template flake by running nix flake init --template github:nix-community/poetry2nix

  4. in the flake.nix, sometimes changing projectDir = self to projectDir = ./. fixed some issues

  5. in your terminal, run nix develop . to build the poetry app with python packages described in pyproject.toml

  6. By default, just poetry and python latest should be installed. the dependencies for the project (which gets reflected in the pyproject.toml) are updated with poetry add, such as poetry add numpy selenium scikit-learn

  7. Exit out of the ephemeral shell from nix develop ., and rerun to have poetry2nix rebuild and link the newly declared packages

Poetry2nix has worked pretty well for the more obscure python packages, but failed in others. For example, sentence-transformers would depend on maturin, which would fail to link setuptools. If poetry doesn't work, you can try and get the package from nixpkgs, or specify sha256s from pypi.org

Here's an example of what I added to my flake.nix to get gpu acceleration, sentence-transfomers, firefox drivers for selenium, and other packages poetry failed to setup:

packages = [ pkgs.poetry pkgs.python311Packages.sentence-transformers pkgs.firefox 
            pkgs.python311Packages.openai pkgs.python311Packages.yt-dlp pkgs.python311Packages.pyopencl
];

was added to this flake.nix, as in,

{
  description = "Application packaged using poetry2nix";

  inputs = {
    flake-utils.url = "github:numtide/flake-utils";
    nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
    poetry2nix = {
      url = "github:nix-community/poetry2nix";
      inputs.nixpkgs.follows = "nixpkgs";
    };
  };
  outputs = { self, nixpkgs, flake-utils, poetry2nix }:
    flake-utils.lib.eachDefaultSystem (system:
      let
        # see https://github.com/nix-community/poetry2nix/tree/master#api for more functions and examples.
        pkgs = nixpkgs.legacyPackages.${system};
        inherit (poetry2nix.lib.mkPoetry2Nix { inherit pkgs; }) mkPoetryApplication;
      in
      {
        packages = {
          myapp = mkPoetryApplication {
            projectDir = ./.;
          };
          default = self.packages.${system}.myapp;
        };
        devShells.default = pkgs.mkShell {
          inputsFrom = [ self.packages.${system}.myapp ];
          packages = [ pkgs.poetry pkgs.python311Packages.sentence-transformers pkgs.firefox 
            pkgs.python311Packages.openai pkgs.python311Packages.yt-dlp pkgs.python311Packages.pyopencl
          ];
          nativeBuildInputs = [(
            pkgs.python311Packages.buildPythonPackage rec {
              pname = "serpapi";
              version = "0.1.5";
              src = pkgs.python311Packages.fetchPypi {
                inherit pname version;
                sha256 = "b9707ed54750fdd2f62dc3a17c6a3fb7fa421dc37902fd65b2263c0ac765a1a5";
              };
            }
          )];
        };
      });
}

There was one package (serpapi), which was not in nixpkgs, and poetry failed as well. Adding this to native build inputs got serpapi installed

nativeBuildInputs = [(
            pkgs.python311Packages.buildPythonPackage rec {
              pname = "serpapi";
              version = "0.1.5";
              src = pkgs.python311Packages.fetchPypi {
                inherit pname version;
                sha256 = "b9707ed54750fdd2f62dc3a17c6a3fb7fa421dc37902fd65b2263c0ac765a1a5";
              };
            }
)];

All in all, it works, and I have no doubt I've made a reproducible environment. What attracts me is I've never had an easier time setting up cuda/cudnn/tensorrt/... system drivers have been near effortless, and much faster to setup than on debian. Tools like sentence-transformers and torch default to packages which leverage the GPU.

What pushes me away, is I've had failures in each of the three methods for specifying package dependencies, even though one of the three eventually was the fix for integrating the dependencies into my shell. For now, I'll stick with it, but it's hard for me to suggest to a team we use this in development

 

I setup a next.js project with pkgs.mkshell, and used nix develop to automatically build the project. However, when I leave the shell, the files persist. How should/can(?) I setup my shell.nix so that files in the directory it drops down into are automatically removed when leaving the ephemeral shell?

view more: next ›