Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models

Name: Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models
Uploaded: 2026-09-05T20:31:41.077773+00:00
Duration: 511 s
Description: The episode discusses 'Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models.' Hosts examine how this method allows researchers to pinpoint which specific component (expert) within a large language model is responsible for factual knowledge. They conclude that understanding these internal mechanics is key to building verifiable and accountable AI systems.

The episode discusses 'Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models.' Hosts examine how this method allows researchers to pinpoint which specific component (expert) within a large language model is responsible for factual knowledge. They conclude that understanding these internal mechanics is key to building verifiable and accountable AI systems.

Open the full episode