Safety Representation Ranking (SRR) trains a lightweight transformer on internal LLM hidden states to rank candidate responses by safety, reporting high pairwise accuracy on safety benchmarks.
Constitutional ai: Harmlessness from ai feedback, 2022
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
-
Advancing LLM Safe Alignment with Safety Representation Ranking
Safety Representation Ranking (SRR) trains a lightweight transformer on internal LLM hidden states to rank candidate responses by safety, reporting high pairwise accuracy on safety benchmarks.