A DPO-based alignment method with a balanced mixture of legal and illegal chemistry prompts improves combined safety and utility scores, but its benchmark shares training compounds and its hyperparameters are tuned on the test set.
Claude 3 haiku: our fastest model yet
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
-
The Dual-use Dilemma in LLMs: Do Empowering Ethical Capacities Make a Degraded Utility?
A DPO-based alignment method with a balanced mixture of legal and illegal chemistry prompts improves combined safety and utility scores, but its benchmark shares training compounds and its hyperparameters are tuned on the test set.