Extracting Concepts from GPT-4

Company news OpenAI Source language: English

Using new techniques for scaling sparse autoencoders, we automatically identified 16 million patterns in GPT-4's computations.

Excerpt

Using new techniques for scaling sparse autoencoders, we automatically identified 16 million patterns in GPT-4's computations.

Short excerpt from the source — open the original for the full text.

Sources

← Back to the daily brief