Robert Lasenby
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- Constitutional AI: Harmlessness from AI FeedbackSaved credit: Contributor · 2022Explore people connected to this work →
- Toy Models of SuperpositionSaved credit: Contributor · 2022Explore people connected to this work →
- Superposition, Memorization, and Double DescentSaved credit: Contributor · 2023Explore people connected to this work →
- Towards Monosemanticity: Decomposing Language Models With Dictionary LearningSaved credit: Contributor · 2023Explore people connected to this work →
- Privileged Bases in the Transformer Residual StreamSaved credit: Contributor · 2023Explore people connected to this work →