I'm more and more confident that tokenization will be gone.
Humans don't think in "tokens".
Tokens are hardcoded abstractions in LLMs that lead to weird behavior: LLMs can solve PhD-level math questions but cannot answer "Is 9.9 > 9.11?"
Meta is shifting LLMs to LCMs (Large
时政
(
twitter.com
)
I'm more and more confident that tokenization will be gone.
Humans don't think in "tokens".
Tokens are hardcoded abstractions in LLMs that lead to weird behavior: LLMs can solve PhD-level math questions but cannot answer "Is 9.9 > 9.11?"
Meta is shifting LLMs to LCMs (Large Concept Models), changing "next-token prediction" to "next-concept prediction". A "concept" is treated as a sentence, representing an abstract idea or action.
Two weeks ago, Meta also published the "Byte Latent Transformer", a new architecture that processes raw bytes grouped into patches instead of relying on tokenization.
My intuition says LCM is probably better at reasoning and modalities.
点击图片查看原图
Today, we can already say there is a good result for our air defense – Ukraine now has more Patriots. I thank Chancellor Merz, I thank Germany, and everyone who helps – our agreements have been fulfilled. More Patriots are now in Ukraine and being put into operation.
Of course,
时政
(
twitter.com)
