ReMix cuts reinforcement finetuning rollout volume by 30x to 450x on math reasoning by mixing historical and on-policy data with a convex KL constraint and a mid-training switch to on-policy updates.
Back to basics: Revisiting reinforce-style optimization for learning from human feedback in llms
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
-
Squeeze the Soaked Sponge: Efficient Off-policy Reinforcement Finetuning for Large Language Model
ReMix cuts reinforcement finetuning rollout volume by 30x to 450x on math reasoning by mixing historical and on-policy data with a convex KL constraint and a mid-training switch to on-policy updates.