Learning a Generative Meta-Model of LLM Activations
- Source
- arxiv.org
- Published
- Feb 6, 2026
Existing approaches for analyzing neural network activations, such as PCA and sparse autoencoders, rely on strong structural assumptions. Generative models offer an alternative: they can uncover structure without such assumptions and act as priors that impr...
arxiv.orgInfo
- Published
- Feb 6, 2026
- Uploaded
- Jul 10, 2026
- Uploaded by
Trevor McFedries
- Queried
- 0 times
Appears in
More
Use with your agent
Have your agent query this content directly
Download package
Unlocks the raw transcripts and files to use as you please
Discover playbooks
Create a repeatable workflow using this source