There's also the performance or let's say the compute you know efficiency advantage because certain things can be run in parallel they're both small. I guess it's compute efficiency and kind of ready parallelization and then third related to the fact that you're predicting bytes you don't have to worry about tokens so you can kind of handle everything as like largely sort of raw data. Yeah so I think for this work it actually comes to hand to hand everything basically but what we really want to solve is get rid ahead of the tokenizer so we want the tokenizer free language model.

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode