We do not need 8.1 Billion ways of writing down the sounds we make and encoding those in symbols.

We do not need 8.1 Billion ways of writing down the sounds we make and encoding those in symbols. Sounds for words, symbols for sounds, a chaos of spellings and fonts. LLM groups choosing arbitrary local tokenizations and not using a finite set of global open tokens – with the cooperation, involvement, agreement and benefit
Read More »