Use Owen-scrambled Sobol points instead of independent Gaussian seeds for batched diffusion sampling, mapping each cube point through the component-wise inverse Gaussian CDF and the model's probability-flow ODE. Estimate ensemble expectations with importance weights computed from the target-to-proposal density ratio, so the estimator remains valid despite finite-step and learned-score transport errors.
Replace unrestricted continuous batching with a fairness-constrained scheduler that only inserts a waiting request when its predicted decode footprint remains within a normalized distance of the longest request already in the batch. The fairness budget should be tuned away from the mathematically worst midpoint and validated against real arrival distributions, with the goal of reducing wasted max-driven decode work without falling back to inefficient one-request-at-a-time serving.
Replace fixed chunked-prefill settings in an LLM serving engine with a feedback controller that decreases the number of prompt tokens processed per scheduling quantum as GPU saturation or long-context load increases. The controller targets a high-quantile bound on the absolute GPU-power ramp while preserving the existing peak-power ceiling and measuring the resulting latency-throughput tradeoff.
Replace a static batch size with a schedule optimized for a prescribed learning-rate schedule and a fixed total number of processed examples. Steps whose stochastic-gradient noise has a large effect on the paper's loss bound receive larger batches, with the weighting determined by the future learning-rate tail rather than by a hand-designed warmup or cooldown rule.