AQI is a latent-activation clustering score intended as a decoding-invariant proxy for LLM alignment quality.
how to secretly override permissions for justice
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Alignment Quality Index (AQI) : Beyond Refusals: AQI as an Intrinsic Alignment Diagnostic via Latent Geometry, Cluster Divergence, and Layer wise Pooled Representations
AQI is a latent-activation clustering score intended as a decoding-invariant proxy for LLM alignment quality.