Tianyi (Alex) Qiu

Gym-Like En­vi­ron­ment for LM Truth-Seeking

Elic­it­ing base mod­els with sim­ple un­su­per­vised techniques

Defin­ing AI Truth-Seek­ing by What It Is Not

• Tianyi (Alex) Qiu 20 Feb 2024 13:29 UTC

  1 point

  0

  on: No Click­bait—Misal­ign­ment Database

  There’s also the goal misgeneralization database by DeepMind, in parallel to the misspecification one: blogpost, database.