“Are Emergent Abilities of Large Language Models a Mirage?”
This is an interesting preprint arguing that the so-called “emergent” abilities of LLMs are, basically, benchmarking artifacts. So much of what’s wrong in AI is down to crap benchmarks
... works as a web developer in Hveragerði, Iceland, and writes about the web, digital publishing, and web/product development
These are his notes
“Are Emergent Abilities of Large Language Models a Mirage?”
This is an interesting preprint arguing that the so-called “emergent” abilities of LLMs are, basically, benchmarking artifacts. So much of what’s wrong in AI is down to crap benchmarks