Topic · 1 episode across 1 review

Building Deletion Into the Model

← all reviews

What if removing a data source from a trained model were a switch you flip rather than a months-long retraining project — and the content were provably gone, not just hidden?

Covered in these reviews

Related concepts

Episodes and papers for this topic live on the concept pages.

Transformer Attention Mechanistic Interpretability AI Safety Evaluation & Benchmarks Training Methods Reward Hacking Post-Training Knowledge Distillation Long Context AI Governance Emergent Behavior