Definition. Dynamic sampling [yu2025dapo, Section 3.2] [ftip-004I]
Definition. Dynamic sampling [yu2025dapo, Section 3.2] [ftip-004I]
DAPO dynamic sampling filters prompt groups whose sampled responses all receive the same reward and continues sampling until the training batch contains the declared number of groups with nonzero reward variation.
The intervention changes the distribution of prompts and rollouts entering the update. It can prevent zero-advantage groups from consuming an optimizer batch, but the retained batch is no longer an unconditioned sample from the original prompt law.