The hidden price tag of refining artificial intelligence
In short: Training a model is only half the battle. This conversation explores the immense resources required for post-training, where raw models are sculpted into useful, safe, and capable AI agents.
Machine learning researchers Nathan Lambert and Sebastian Raschka discuss the complexities of post-training—the phase following initial pre-training. While pre-training establishes a model's base knowledge, post-training is what allows it to follow instructions, reason effectively, and align with human values.
The discussion highlights the roles of experts from the Allen Institute for AI and authors of foundational texts like 'The RLHF Book.' It examines the technical and financial hurdles of moving from a raw large language model to a functional product, emphasizing why this stage is critical for modern AI deployment.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: How costly is post-training of LLMs | Lex Fridman Podcast
To cite this card, link to https://intellectplus.co.uk/app/science-how-costly-is-post-training-of-llms-lex-fridman-podcast.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Science · Latest · A–Z · All topics
