Skip Navigation

Kevin

@ technohacker @programming.dev

Posts
1
Comments
91
Joined
3 yr. ago

  • lsfg-vk aims to bring Lossless Scaling's Frame Generation to Linux

    Jump

  • But as far as I understand, it's designed to be a Vulkan layer. I'm pretty sure they can be installed under the user's home directory at ~/.local and picked up by Mesa automatically

  • ..........................

    Jump

  • Oh I just noticed the AI artifacts, the powerlines and license plate in particular

  • Moonbase Alpha: That time NASA made a meme video game

    Jump

  • Football!

  • A random thought while hanging laundry

    Jump

  • Oddly enough your post reminded me to take out my laundry from the washing machine. Thanks!

  • Don't do Caffeine and Complex Analysis, kids

    Jump

  • I dunno if I wanna have 8 glasses of vegetables

  • Outstanding work

    Jump

  • Is this...?

  • based on a real situation

    Jump

  • Pivotal Labs originally, then joined VMWare Tanzu I think, Now in turn bought by Broadcom

  • Consciousness

    Jump

  • Segmentation Fault (Core Dumped)

  • Starsector, one of the best space RPGs that's not on Steam had a huge update

    Jump

  • Huh, i thought PZ shipped a Java runtime?

  • Starsector, one of the best space RPGs that's not on Steam had a huge update

    Jump

  • Project Zomboid too AFAIK

  • That explains a lot

    Jump

  • Foreign Music

    Jump

  • Moksori's playlists on YouTube, really great collections

  • pro tip

    Jump

  • Oof right in the feels

  • Dirty Blood Sucking Metasites

    Jump

  • PyTorch is now under the Linux Foundation AFAIK, but yes it did originate at FB

  • Scanning the archives

    Jump

  • Ow i feel called out

  • Nothing I swear!!

    Jump

  • Goose Willis

  • Teabags

    Jump

  • iykyk

    Jump

  • (1,2,2,50)-loss-quinquagintinane

  • If we had libre AI

    Jump

  • I would actually bring a parallel to the device driver-firmware blob split that's common with hardware support in Linux. While the code needed to run inference with a model is straightforward and several open source versions exist already, the model itself is a bunch of tensors whose behaviour we don't have any visibility into. Bias is less a problem of the inference code and more an issue with the data it was trained on