Results for "weighted sampling"

Stochastic generation strategies that trade determinism for diversity; key knobs include temperature and nucleus sampling.

Sampling from easier distribution with reweighting.

Selecting the most informative samples to label (e.g., uncertainty sampling) to reduce labeling cost.

Scales logits before sampling; higher increases randomness/diversity, lower increases determinism.

Variability introduced by minibatch sampling during SGD.

Learns the score (∇ log p(x)) for generative sampling.

Approximating expectations via random sampling.

Sampling multiple outputs and selecting consensus.

Sampling-based motion planner.