#训练-推理不一致

1 stories · 0 in 7 days · 0 in 30 days

Stories, products and papers tagged "训练-推理不一致", newest first.

Sep 18

Score Centering Stabilizes Off-policy RL

For researchers working on reinforcement learning, this paper presents a practical method to stabilize RL under training-inference mismatch, which is a common challenge when scaling models.

#reinforcement learning #off-policy reinforcement learning #training-inference mismatch #RL algorithms

arXiv cs.LG@Martin Marek, Max RyabininSource