• crysisnotaverted@alien.topB
    link
    fedilink
    English
    arrow-up
    3
    arrow-down
    1
    ·
    11 months ago

    I might be too dumb for this one, boys. I can’t wrap my head around what “Open source AI Agent evaluations for web tasks” means…

    Other than me being stupid, that is one well designed github repo, lol.

    • asim-shrestha@alien.topOPB
      link
      fedilink
      English
      arrow-up
      1
      ·
      11 months ago

      😂 a bit opaque if you’re not super familiar with the space i suppose

      ELI5: Theres a lot of work being done with LLMs to take actions on websites. This open source repo provides static versions of these websites along with some evaluation criteria to measure the performance of your LLM “agents”. Its quite a pain to reliably test these agents otherwise. (An agent being some system of code that will take a goal like “travel to xyz on this page” and use an llm to translate that into actual actions)