NEW Multi-Agent Dynamics w/ Self-Play RL

Поділитися
Вставка
  • Опубліковано 17 гру 2024

КОМЕНТАРІ • 4

  • @themax2go
    @themax2go 4 місяці тому +2

    dang i should've written the paper when i had the idea... i've been thinking of creating that w/ internet search added, kinda a "self-running, self-motivated, chat-with-each-other team / agents" - of course with long-term memory implemented via triplets / sci-phi triplex, to also be able to develop and pull global contexts and effectively cross-reference between all "memories"... kinda like a "borg hive" (as in, the fictional faction / civ from star trek) with 1 long-term memory but many brains (short-term mem and individual compute capability)

  • @vitalis
    @vitalis 4 місяці тому +1

    “Self-play” 🗿😎

  • @antoniotonino2909
    @antoniotonino2909 4 місяці тому

    I see the cost in tokens before the actual task is growing.

  • @pappaflammyboi5799
    @pappaflammyboi5799 4 місяці тому +1

    We should rename it "The Borg".