SFT-trained reward models over 12 dialogue-impression metrics, used as AI feedback in DPO, improve Japanese chit-chat systems' naturalness and impression scores over no tuning.
An overview of bard: an early experiment with gener- ative ai, 2023
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Training Dialogue Systems by AI Feedback for Improving Overall Dialogue Impression
SFT-trained reward models over 12 dialogue-impression metrics, used as AI feedback in DPO, improve Japanese chit-chat systems' naturalness and impression scores over no tuning.