The transformer architecture has not significantly changed since its inception, except for the introduction of the mixture of experts model. While concerns exist about quadratic attention complexities, addressing them may lead to increased costs in matrix multiplies. The majority of computation in large transformers like GPT-3 or GPT-4 is spent on matrix multiplications, not the attention layer. Therefore, innovations in auxiliary layers like retrieval and augmented generation may hold more potential than focusing on optimizing the transformer architecture itself for models like GPT-4 due to the high costs associated with training such large models.

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