Gemma 2-2B research finds active feature planes have less holonomy

A new research paper published on arXiv investigates the concentration of holonomy within specific feature planes of the Gemma 2-2B model. The study preregistered its methodology and analysis rules before inspecting the data. Contrary to the prediction that active feature planes would exhibit more holonomy, the results showed the opposite, with active planes carrying less holonomy than control groups. The paper concludes this is an auditable operational reversal rather than a definitive causal claim, leaving the underlying cause open to further investigation. AI

IMPACT This research offers a novel perspective on internal model mechanics, potentially influencing future interpretability techniques.

RANK_REASON The cluster contains a research paper detailing an experiment and findings related to a specific AI model. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gemma
• Gemma 2-2B
• Gotit.pub
• Hugging Face
• IArxiv
• Jacobian matrix
• SAE International
• ScienceCast

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →