Berkeley AI Research·· 2026-03-13
Identifying Interactions at Scale for LLMs
Identifying Interactions at Scale for LLMs
AI summary
Berkeley AI Research proposed SPEX and ProxySPEX to identify key interactions driving LLM outputs at scale in feature, data and model-component attribution. SPEX turns interaction search into sparse recovery using sparsity and low order; ProxySPEX exploits hierarchy to match SPEX with roughly ten times fewer ablations.
Selection record
Threshold 60Official, first-handFirst 34Second 31
Not admittedSum of both 65 < twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：LLM可解释性交互发现算法研究
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Berkeley AI Research · bair.berkeley.edu