New Deepseek model drastically reduces resource usage by converting text and documents into images — 'vision-text compression' uses up to 20 times fewer tokens

Comments

• yakko@feddit.uk ⁨1⁩ ⁨week⁩ ago

  I feel like you never see anyone use maths language correctly anymore. The word is “one twentieth”.

  source
• Arcane2077@sh.itjust.works ⁨1⁩ ⁨week⁩ ago

  When did images become lighter than text??

  source

  • technocrit@lemmy.dbzer0.com ⁨1⁩ ⁨week⁩ ago

    I thought the same thing.

    It’s an “AI” story so they bury the actual result behind a sensational headline:

    This works really well for handling tabulated data, graphs, and other visual representations of information.

    source
  • Evotech@lemmy.world ⁨1⁩ ⁨week⁩ ago

    A picture is worth a thousand words /s

    source