Standard fine-tuning of user simulators injects look-ahead bias that causes controllability collapse under policy shift, which is fixed by a priori, step-wise, or policy-conditioned controls.
* ‘skill_level‘: The apparent expertise of the user regarding the topic (e.g., beginner coder, fanfiction enthusiast, student, general public)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Controllable User Simulation
Standard fine-tuning of user simulators injects look-ahead bias that causes controllability collapse under policy shift, which is fixed by a priori, step-wise, or policy-conditioned controls.