Earlier quoted context omitted.
All it speaks to is that tokenization is weird and introduces artifacts to LLM performance. Counting letters is a trivial task when you're staring at words on a screen. It's much harder when you're perceiving vectors based on parts of words. The fact that LLMs find certain things easier/harder than humans is completely unsurprising, and there are much more interesting benchmarks to use to compare one LLM to another.
This is a good point. While LLMs being incapable of reliably doing a simple task that’s been doable by computers since the punch card days is an important consideration for anyone that might be thinking about using them for anything other than as a toy , this fact is uninteresting because of Reasons
Don’t need to be exact as firing is statistical, just give us a good average.
Hard? You can’t count?
Computers count pixels no problem. So weird you can’t.
Dementia? Not an AGI? /h
—-
This is what is happening.
Here are the “Reasons”.
In your vision system, the raw information from individual retina signals is munged into a different representation before reaching a level where you have flexible processing.
Likewise, in LLMs, letters are munged into tokens before LLMs “see” them.
When they sometimes get that “simple” question right, it’s actually a bit of an amazing feat. Given how they are constructed.
—-
Now try counting R’s as you read at a normal rate, or listen to someone speak.
You can’t do that either, during normal processing.
When we add spelling to LLMs training examples, they will do it easily. Just as you learned to do it, only after special lessons, after you had already learned to listen and speak.
Spelling is its own special practiced skill, in humans and LLMs.