SITAlign is an inference-time constrained decoder that maximizes a primary reward while enforcing thresholds on secondary rewards, and it reports better primary-reward win-tie rates than weighted-objective decoding.
Safe rlhf: Safe reinforcement learning from human feedback
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Bounded Rationality for LLMs: Satisficing Alignment at Inference-Time
SITAlign is an inference-time constrained decoder that maximizes a primary reward while enforcing thresholds on secondary rewards, and it reports better primary-reward win-tie rates than weighted-objective decoding.