RoleConflictBench generates role-conflict scenarios by varying situational urgency to measure whether LLMs follow dynamic context or learned role preferences, finding that tested models mostly follow the latter.
Here is the description of 10 values and their underlying motivators
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
RoleConflictBench: A Benchmark of Role Conflict Scenarios for Evaluating LLMs' Contextual Sensitivity
RoleConflictBench generates role-conflict scenarios by varying situational urgency to measure whether LLMs follow dynamic context or learned role preferences, finding that tested models mostly follow the latter.