We're using kind of the same encoding in that experiment and finding that when you're dealing with bytes like the megabyte architecture just blows away the others. The key punchline is the results are comparable yeah. We implement this perceiver AR in house of course we run the experiments to validate we get comparable experiments from original paper but in a compute experiment we run the baseline which is perceiver AR of course and another baseline which is naive transformer.

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