DRIFT achieves multi-turn RL performance via offline importance-weighted SFT by leveraging the equivalence of KL-regularized RL to weighted supervised learning.
On- line preference alignment for language models via count- based exploration.arXiv preprint arXiv:2501.12735,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
DRIFT: Decoupled Rollouts and Importance-Weighted Fine-Tuning for Efficient Multi-Turn Optimization
DRIFT achieves multi-turn RL performance via offline importance-weighted SFT by leveraging the equivalence of KL-regularized RL to weighted supervised learning.