Reasoning-enhancing prompts and fine-tuning, especially long chain-of-thought data, tend to reduce LLM safety scores, and mixing in reflective safety data only partially restores them.
Training verifiers to solve math word problems
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
dataset 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
dataset 1polarities
use dataset 1representative citing papers
citing papers explorer
-
Are Smarter LLMs Safer? Exploring Safety-Reasoning Trade-offs in Prompting and Fine-Tuning
Reasoning-enhancing prompts and fine-tuning, especially long chain-of-thought data, tend to reduce LLM safety scores, and mixing in reflective safety data only partially restores them.