Latent Space: The AI Engineer Podcast cover image

AI Magic: Shipping 1000s of successful products with no managers and a team of 12 — Jeremy Howard of Answer.ai

Latent Space: The AI Engineer Podcast

NOTE

Merge Wisely: Embrace Adapters Over Models

Merging models can yield interesting results, but it's essential to adopt a cautious approach. Instead of merging models directly, it’s better to distribute merged adapters, which have been shown to perform effectively. This method allows users to leverage a quantized model they've already downloaded, resulting in a smaller download size. Moreover, this strategy enhances inference speed, as there's no need to transfer heavy FB16 weights. By focusing on adapters, both inference and fine-tuning processes can be expedited, showcasing the benefits of a well-structured model merging practice.

00:00
Transcript
Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner