INTERVIEW: Polysemanticity w/ Dr. Darryl Wright
From Into AI Safety · 45:09
Show by Jacob Haimes
Darryl and I discuss his background, how he became interested in machine learning, and a project we are currently working on investigating the penalization of polysemanticity during the training of neural networks.Check out a diagram of the decoder task used for our research!01:46 - Interview begins02:14 - Supernovae classification08:58 - Penalizing polysemanticity20:58 - Our "toy model"30:06 - Task description32:47 - Addressing hurdles39:20 - Lessons learnedLinks to all articles/papers which are mentioned throughout the episode can be found below, in order of their appearance.ZooniverseBlueDot ImpactAI Safety SupportZoom In: An Introduction to CircuitsMNIST dataset on PapersWithCodeClusterability in Neural NetworksCIFAR-10 datasetEffective Altruism GlobalCLIP (blog post)Long Term Future FundEngineering Monosemanticity in Toy Models
More episodes of Into AI Safety
More episodes
- Unbrokered Agent Coordination w/ Alex Obadia
- Holding the Human Line w/ Allan Brooks and Etienne Brisson
- MINISODE: Introduction and Motivation
- FEEDBACK: AISC Proposal w/ Remmelt Ellen
- MINISODE: EAG Takeaways (Boston 2023)
- MINISODE: Program Applications (Winter 2024)
- INTERVIEW: Applications w/ Alice Rigg
- MINISODE: Staying Up-to-Date in AI
- HACKATHON: Evals November 2023 (1)
- MINISODE: Starting a Podcast
PodSearch is a free, independent podcast search engine. Read about PodSearch, the privacy policy or contact us.