Re-distilling a model's own RL-trained policy into 1K SFT samples reproduces RL accuracy at a fraction of the compute.
Quantifying memorization across neural language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Revealing the Effectiveness of Small-Scale Fine-tuning in R1-style Reinforcement Learning
Re-distilling a model's own RL-trained policy into 1K SFT samples reproduces RL accuracy at a fraction of the compute.