Training an open source model to mimic GPT-4 reasoning has become a challenge. While some models achieve good scores on benchmarks, they still fall short of GPT-4's capabilities. Spending more compute at runtime could be a game-changer. Gemini may be exploring this avenue. The finishing technique is powerful, but there is potential for more gains by asking different questions and using tools in innovative ways. Breaking down questions into smaller pieces and building them back up leads to better results, as seen in novel writing tools.

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