‹ BackHN Continuity

Thread

Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data

158 points · 43 comments · Betelbuddy

  1. alightsoul · · focus · HN ↗
    So would it be a 42 trillion parameter model, because that's how many tokens there are in the training data?
    1. k__ · · focus · HN ↗
      Does that mean, it's not compressed anymore?
    2. robotresearcher · · focus · HN ↗
      No, the parameters are updated online as new run-time data comes in. Infinite different param values over time, but a constant number of params. It's a stretch really, since by this usage a one-parameter model that updates itself is also infinite over time...
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.