Transformer architecture is native for this and it works really well. We make prediction every token so basically if you have 16 tokens you make 16 predictions in the auto-regressive model. The parallelism of the transformer we're able to during training during training you can do the feed forward and get every loss  to power your back propagation optimization that's presumably particularly useful for like informing the global model but once you're past the global model now there's no more communication right between the local models, he says.

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