This raises an obvious question,how should the cost-tradeoffs be chosen so that it reflects the expert'sdecision-making?

Large language model (LLM) inference workload dominates a wide variety of modern AI applications, ranging from multi-turn conversation to document analysis. Balancing fairness and efficiency is critical for managing diverse client workloads with varying prefix patterns. Unfortunately, existing fair scheduling algorithms for LLM serving, such as Virtual Token Counter (VTC), fail to take prefix locality into consideration and thus suffer from poor performance. On the other hand, locality-aware scheduling algorithms in existing LLM serving frameworks tend to maximize the prefix cache hit rate without considering fair sharing among clients. This paper introduces the first locality-aware fair scheduling algorithm, Deficit Longest Prefix Match (DLPM), which can maintain a high degree of prefix locality with a fairness guarantee. We also introduce a novel algorithm, Double Deficit LPM (D$^2$LPM), extending DLPM for the distributed setup that can find a balance point among fairness, locality, and load-balancing. Our extensive evaluation demonstrates the superior performance of DLPM and D$^2$LPM in ensuring fairness while maintaining high throughput (up to 2.87$\times$ higher than VTC) and low per-client (up to 7.18$\times$ lower than state-of-the-art distributed LLM serving system) latency.

This paper studies the metric elicitation problem, a problem proposed by [7] (AISTATS'19 paper). The authors defined two types of metrics for multi-class classification, Diagonal Linear Performance Metric (DLPM) and Linear Performance Metric (LPM), and designed algorithms for the two type of metrics respectively. The algorithm for DLPM is relatively simple: just use binary-search like [7] and apply it for (k-1) times. The LPM case is much more complicated - the authors proposed a coordinate-wise binary-search style algorithm based on the geometry of the feasible set. Theoretical guarantees for both algorithms are provided.

Investigating noise distribution beyond Gaussian in diffusion generative models is an open problem. The Gaussian case has seen success experimentally and theoretically, fitting a unified SDE framework for score-based and denoising formulations. Recent studies suggest heavy-tailed noise distributions can address mode collapse and manage datasets with class imbalance, heavy tails, or outliers. Yoon et al. (NeurIPS 2023) introduced the L\'evy-Ito model (LIM), extending the SDE framework to heavy-tailed SDEs with $\alpha$-stable noise. Despite its theoretical elegance and performance gains, LIM's complex mathematics may limit its accessibility and broader adoption. This study takes a simpler approach by extending the denoising diffusion probabilistic model (DDPM) with $\alpha$-stable noise, creating the denoising L\'evy probabilistic model (DLPM). Using elementary proof techniques, we show DLPM reduces to running vanilla DDPM with minimal changes, allowing the use of existing implementations with minimal changes. DLPM and LIM have different training algorithms and, unlike the Gaussian case, they admit different backward processes and sampling algorithms. Our experiments demonstrate that DLPM achieves better coverage of data distribution tail, improved generation of unbalanced datasets, and faster computation times with fewer backward steps.