Hacker News new | past | comments | ask | show | jobs | submit
jygg4 | on: Are LLM merge rates not getting better?
The models lose the ability to inject subtle and nuance stuff as they scale up, is what I’ve observed.